pr paths node_dis_index pr_dis para commentary commit_s commit_e draft para_hash block_hash url AB paragraph mail_commentary 220 {} {} {} "In order to maximize HTTP cache friendliness, DoH clients using media formats that include DNS ID, such as application/dns-message, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as application/dns-message. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." These look good to me . 74800232507ad01df6c1c5b525d7aab84ef91759 74800232507ad01df6c1c5b525d7aab84ef91759 draft-ietf-doh-dns-over-https-latest 83131119fbae4a2a51803cd4d51d04e82174bb95966e5287713b0eec6378542b ['83131119fbae4a2a51803cd4d51d04e82174bb95966e5287713b0eec6378542b', '6e9cb9573765803abbf772c95319023018994095d1f5bdfad871c82fd62042f6'] A "in order to maximize http cache friendliness, doh clients using media formats that include dns id, such as application/dns-message, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as application/dns-message. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 220 {} {} {} "In order to maximize HTTP cache friendliness, DoH clients using media formats that include the ID field from the DNS message header, such as application/dns-message, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as application/dns-message. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." These look good to me . 74800232507ad01df6c1c5b525d7aab84ef91759 74800232507ad01df6c1c5b525d7aab84ef91759 draft-ietf-doh-dns-over-https-latest 6e9cb9573765803abbf772c95319023018994095d1f5bdfad871c82fd62042f6 ['83131119fbae4a2a51803cd4d51d04e82174bb95966e5287713b0eec6378542b', '6e9cb9573765803abbf772c95319023018994095d1f5bdfad871c82fd62042f6'] B "in order to maximize http cache friendliness, doh clients using media formats that include the id field from the dns message header, such as application/dns-message, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as application/dns-message. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 220 {} {} {} "Some HTTPS client implementations perform real time third-party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DoH server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of OCSP RFC6960 servers or AIA for CRL fetching (RFC5280 Section 4.2.2.1) are examples of how this deadlock can happen. To mitigate the possibility of deadlock, DoH servers SHOULD NOT rely on DNS-based references to external resources in the TLS handshake. For OCSP, the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC8446 Section 4.4.2.1 for TLS version 1.3. AIA deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. Note that these deadlocks also need to be considered for servers that a DoH server might redirect to." These look good to me . 74800232507ad01df6c1c5b525d7aab84ef91759 74800232507ad01df6c1c5b525d7aab84ef91759 draft-ietf-doh-dns-over-https-latest f1a062701f986b72b07ec8e1ef83c9715b1a08f75e41a80c7a9c313c2b9ac85d ['cf6ee1ae8e7b863accdfc04ad6078f0dcfe7ad1f50289e5f0002eb51100e0639', 'f1a062701f986b72b07ec8e1ef83c9715b1a08f75e41a80c7a9c313c2b9ac85d'] A "some https client implementations perform real time third-party checks of the revocation status of the certificates being used by tls. if this check is done as part of the doh server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of ocsp rfc6960 servers or aia for crl fetching (rfc5280 section 4.2.2.1) are examples of how this deadlock can happen. to mitigate the possibility of deadlock, doh servers should not rely on dns-based references to external resources in the tls handshake. for ocsp, the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc8446 section 4.4.2.1 for tls version 1.3. aia deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. note that these deadlocks also need to be considered for servers that a doh server might redirect to." 220 {} {} {} "Some HTTPS client implementations perform real time third-party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DoH server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of OCSP RFC6960 servers or AIA for CRL fetching (RFC5280 Section 4.2.2.1) are examples of how this deadlock can happen. To mitigate the possibility of deadlock, the authentication given DoH servers SHOULD NOT rely on DNS-based references to external resources in the TLS handshake. For OCSP, the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC8446 Section 4.4.2.1 for TLS version 1.3. AIA deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. Note that these deadlocks also need to be considered for servers that a DoH server might redirect to." These look good to me . 74800232507ad01df6c1c5b525d7aab84ef91759 74800232507ad01df6c1c5b525d7aab84ef91759 draft-ietf-doh-dns-over-https-latest cf6ee1ae8e7b863accdfc04ad6078f0dcfe7ad1f50289e5f0002eb51100e0639 ['cf6ee1ae8e7b863accdfc04ad6078f0dcfe7ad1f50289e5f0002eb51100e0639', 'f1a062701f986b72b07ec8e1ef83c9715b1a08f75e41a80c7a9c313c2b9ac85d'] B "some https client implementations perform real time third-party checks of the revocation status of the certificates being used by tls. if this check is done as part of the doh server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of ocsp rfc6960 servers or aia for crl fetching (rfc5280 section 4.2.2.1) are examples of how this deadlock can happen. to mitigate the possibility of deadlock, the authentication given doh servers should not rely on dns-based references to external resources in the tls handshake. for ocsp, the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc8446 section 4.4.2.1 for tls version 1.3. aia deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. note that these deadlocks also need to be considered for servers that a doh server might redirect to." 210 {37, 79} {37: 0, 79: 1} {210: 0, 44: 1} "This is an example response for a query for the IN A records for ""www.example.com"" with recursion turned on. The response bears one record with an address of 192.0.2.1 and a TTL of 128 seconds." … 3849 c1ea4fb51c01df0e70a28b382cfd840639bbbf56 c1ea4fb51c01df0e70a28b382cfd840639bbbf56 draft-ietf-doh-dns-over-https-latest aab29aa5a84ed9e4b43de18963d935d378768f40bd644a678edfd8d929a7b8c3 ['56b07e09fc117f916365fed31303943b9696426dde3304d308f9f4f11ec81126', 'aab29aa5a84ed9e4b43de18963d935d378768f40bd644a678edfd8d929a7b8c3'] A "this is an example response for a query for the in a records for ""www.example.com"" with recursion turned on. the response bears one record with an address of 192.0.2.1 and a ttl of 128 seconds." 210 {37, 79} {37: 0, 79: 1} {210: 0, 44: 1} "This is an example response for a query for the IN AAAA records for ""www.example.com"" with recursion turned on. The response bears one answer record with an address of 2001:db8:abcd:12:1:2:3:4 and a TTL of 3709 seconds." … 3849 c1ea4fb51c01df0e70a28b382cfd840639bbbf56 c1ea4fb51c01df0e70a28b382cfd840639bbbf56 draft-ietf-doh-dns-over-https-latest 56b07e09fc117f916365fed31303943b9696426dde3304d308f9f4f11ec81126 ['56b07e09fc117f916365fed31303943b9696426dde3304d308f9f4f11ec81126', 'aab29aa5a84ed9e4b43de18963d935d378768f40bd644a678edfd8d929a7b8c3'] B "this is an example response for a query for the in aaaa records for ""www.example.com"" with recursion turned on. the response bears one answer record with an address of 2001:db8:abcd:12:1:2:3:4 and a ttl of 3709 seconds." 207 {} {} {} "Session-level encryption has well-known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (RFC7540 Section 10.6) and padding (RFC7540 Section 10.7 ). DoH Servers can also add DNS padding RFC7830 if the DoH client requests it in the DNS query." on padding length which was absent from RFC 7830 . 82b9d676cec8cab1ad32cbc6207bee9a8369467b 82b9d676cec8cab1ad32cbc6207bee9a8369467b draft-ietf-doh-dns-over-https-latest 7dbc1f4528565e4dfe69a10ef6b8f77b414687473d5c75187d00220314695288 ['62f298e32cb560f96375312be9fca8a4ee9f8f236717fccb66b7885678179e77', '7dbc1f4528565e4dfe69a10ef6b8f77b414687473d5c75187d00220314695288'] A "session-level encryption has well-known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (rfc7540 section 10.6) and padding (rfc7540 section 10.7 ). doh servers can also add dns padding rfc7830 if the doh client requests it in the dns query." 207 {} {} {} "Session-level encryption has well-known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (RFC7540 Section 10.6) and padding (RFC7540 Section 10.7 ). DoH Servers can also add DNS padding RFC7830 if the DoH client requests it in the DNS query. Initial attempts to offer guidance on choosing the padding length can be found in draft-ietf-dprive- padding-policy." on padding length which was absent from RFC 7830 . 82b9d676cec8cab1ad32cbc6207bee9a8369467b 82b9d676cec8cab1ad32cbc6207bee9a8369467b draft-ietf-doh-dns-over-https-latest 62f298e32cb560f96375312be9fca8a4ee9f8f236717fccb66b7885678179e77 ['62f298e32cb560f96375312be9fca8a4ee9f8f236717fccb66b7885678179e77', '7dbc1f4528565e4dfe69a10ef6b8f77b414687473d5c75187d00220314695288'] B "session-level encryption has well-known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (rfc7540 section 10.6) and padding (rfc7540 section 10.7 ). doh servers can also add dns padding rfc7830 if the doh client requests it in the dns query. initial attempts to offer guidance on choosing the padding length can be found in draft-ietf-dprive- padding-policy." 203 {42, 109, 15, 52, 86, 95} {42: 0, 52: 1, 15: 1, 109: 2, 86: 2, 95: 2} {203: 0, 200: 1, 202: 1, 201: 2} "DoH encrypts DNS traffic and requires authentication of the server. This mitigates both passive surveillance RFC7258 and active attacks attempting to divert DNS traffic to rogue servers (RFC7626 Section 2.5.1). DNS over TLS RFC7858 provides similar protections, while direct UDP and TCP based transports are vulnerable to this class of information leak." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest 66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19 ['66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', '1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68'] A "doh encrypts dns traffic and requires authentication of the server. this mitigates both passive surveillance rfc7258 and active attacks attempting to divert dns traffic to rogue servers (rfc7626 section 2.5.1). dns over tls rfc7858 provides similar protections, while direct udp and tcp based transports are vulnerable to this class of information leak." 203 {42, 109, 15, 52, 86, 95} {42: 0, 52: 1, 15: 1, 109: 2, 86: 2, 95: 2} {203: 0, 200: 1, 202: 1, 201: 2} "Additionally, the use of the HTTPS default port 443 and the ability to mix DoH traffic with other HTTPS traffic on the same connection can deter on-path devices from interfering with DNS operations and make DNS traffic analysis more difficult." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c ['66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', '1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68'] A "additionally, the use of the https default port 443 and the ability to mix doh traffic with other https traffic on the same connection can deter on-path devices from interfering with dns operations and make dns traffic analysis more difficult." 203 {42, 109, 15, 52, 86, 95} {42: 0, 52: 1, 15: 1, 109: 2, 86: 2, 95: 2} {203: 0, 200: 1, 202: 1, 201: 2} "A DoH implementation is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate different queries to the same identity. DNS transports will generally carry the same privacy properties of the layers used to implement them. For example, the properties of IP, TCP, and TLS apply to DNS over TLS implementations." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest 1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde ['66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', '1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68'] A "a doh implementation is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate different queries to the same identity. dns transports will generally carry the same privacy properties of the layers used to implement them. for example, the properties of ip, tcp, and tls apply to dns over tls implementations." 203 {42, 109, 15, 52, 86, 95} {42: 0, 52: 1, 15: 1, 109: 2, 86: 2, 95: 2} {203: 0, 200: 1, 202: 1, 201: 2} "DoH encrypts DNS traffic and requires authentication of the server. This mitigates both passive surveillance RFC7258 and active attacks that attempt to divert DNS traffic to rogue servers (RFC7626 Section 2.5.1). DNS over TLS RFC7858 provides similar protections, while direct UDP and TCP based transports are vulnerable to this class of attack." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest 8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608 ['66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', '1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68'] B "doh encrypts dns traffic and requires authentication of the server. this mitigates both passive surveillance rfc7258 and active attacks that attempt to divert dns traffic to rogue servers (rfc7626 section 2.5.1). dns over tls rfc7858 provides similar protections, while direct udp and tcp based transports are vulnerable to this class of attack." 203 {42, 109, 15, 52, 86, 95} {42: 0, 52: 1, 15: 1, 109: 2, 86: 2, 95: 2} {203: 0, 200: 1, 202: 1, 201: 2} "Additionally, the use of the HTTPS default port 443 and the ability to mix DoH traffic with other HTTPS traffic on the same connection can deter unprivileged on-path devices from interfering with DNS operations and make DNS traffic analysis more difficult." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b ['66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', '1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68'] B "additionally, the use of the https default port 443 and the ability to mix doh traffic with other https traffic on the same connection can deter unprivileged on-path devices from interfering with dns operations and make dns traffic analysis more difficult." 203 {42, 109, 15, 52, 86, 95} {42: 0, 52: 1, 15: 1, 109: 2, 86: 2, 95: 2} {203: 0, 200: 1, 202: 1, 201: 2} "A DoH implementation is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate queries to the same identity. DNS transports will generally carry the same privacy properties of the layers used to implement them. For example, the properties of IP, TCP, and TLS apply to DNS over TLS implementations." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest 8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68 ['66c2923e3c778f485334b1d4d80653233e6ec5aaf132fa150568e8be6cb7dd19', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', '1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'afa4574f57baa96f5a96e20ef5ad5251f8f14580e8b03dd662a8eee1112bcb2c', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68'] B "a doh implementation is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate queries to the same identity. dns transports will generally carry the same privacy properties of the layers used to implement them. for example, the properties of ip, tcp, and tls apply to dns over tls implementations." 203 {42, 109, 15, 52, 86, 95} {95: 0, 52: 1, 42: 2, 109: 2, 86: 2, 15: 3} {203: 0, 200: 1, 201: 2, 202: 2} "TCP-based solutions may seek performance through the use of TCP Fast Open RFC7413. The cookies used in TCP Fast Open allow servers to correlate different TCP sessions together." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest 4cb5ffbcf77fb95da3f0fcff687c8bc017cf4ea3b3b2b7b3814e6cdd7eefddcc ['cbe48907e5a954dc0cc3e9bcf08564990eca3b94b59a62c6bf39cec0688d930f', '76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '4cb5ffbcf77fb95da3f0fcff687c8bc017cf4ea3b3b2b7b3814e6cdd7eefddcc'] A "tcp-based solutions may seek performance through the use of tcp fast open rfc7413. the cookies used in tcp fast open allow servers to correlate different tcp sessions together." 203 {42, 109, 15, 52, 86, 95} {95: 0, 52: 1, 42: 2, 109: 2, 86: 2, 15: 3} {203: 0, 200: 1, 201: 2, 202: 2} "TLS based implementations often achieve better handshake performance through the use of some form of session resumption mechanism such as session tickets RFC5077. Session resumption creates trivial mechanisms for a server to correlate different TLS connections together." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest cbe48907e5a954dc0cc3e9bcf08564990eca3b94b59a62c6bf39cec0688d930f ['cbe48907e5a954dc0cc3e9bcf08564990eca3b94b59a62c6bf39cec0688d930f', '76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '4cb5ffbcf77fb95da3f0fcff687c8bc017cf4ea3b3b2b7b3814e6cdd7eefddcc'] A "tls based implementations often achieve better handshake performance through the use of some form of session resumption mechanism such as session tickets rfc5077. session resumption creates trivial mechanisms for a server to correlate different tls connections together." 203 {42, 109, 15, 52, 86, 95} {95: 0, 52: 1, 42: 2, 109: 2, 86: 2, 15: 3} {203: 0, 200: 1, 201: 2, 202: 2} "TCP-based solutions may seek performance through the use of TCP Fast Open RFC7413. The cookies used in TCP Fast Open allow servers to correlate TCP sessions together." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest 76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3 ['cbe48907e5a954dc0cc3e9bcf08564990eca3b94b59a62c6bf39cec0688d930f', '76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '4cb5ffbcf77fb95da3f0fcff687c8bc017cf4ea3b3b2b7b3814e6cdd7eefddcc'] B "tcp-based solutions may seek performance through the use of tcp fast open rfc7413. the cookies used in tcp fast open allow servers to correlate tcp sessions together." 203 {42, 109, 15, 52, 86, 95} {95: 0, 52: 1, 42: 2, 109: 2, 86: 2, 15: 3} {203: 0, 200: 1, 201: 2, 202: 2} "TLS based implementations often achieve better handshake performance through the use of some form of session resumption mechanism such as session tickets RFC5077. Session resumption creates trivial mechanisms for a server to correlate TLS connections together." this is a PR against these are pretty straight forward e3c768e3197d3a524e8fd177f290607aa343074b e3c768e3197d3a524e8fd177f290607aa343074b draft-ietf-doh-dns-over-https-latest e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7 ['cbe48907e5a954dc0cc3e9bcf08564990eca3b94b59a62c6bf39cec0688d930f', '76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '4cb5ffbcf77fb95da3f0fcff687c8bc017cf4ea3b3b2b7b3814e6cdd7eefddcc'] B "tls based implementations often achieve better handshake performance through the use of some form of session resumption mechanism such as session tickets rfc5077. session resumption creates trivial mechanisms for a server to correlate tls connections together." 202 {42, 109, 15, 52, 86, 95} {15: 0, 42: 1, 109: 1, 52: 2, 86: 3, 95: 3} {202: 0, 203: 1, 201: 1, 200: 2} "A DoH application is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate different queries to the same identity. DNS transports will generally carry the same privacy properties of the layers used to implement them. For example, the properties of IP, TCP, and TLS apply to DNS over TLS implementations." this pr is against pr c61b08d0605e451d66ea5538d0e6b1ab13146b9f c61b08d0605e451d66ea5538d0e6b1ab13146b9f draft-ietf-doh-dns-over-https-latest 5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176 ['1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', '5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176'] A "a doh application is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate different queries to the same identity. dns transports will generally carry the same privacy properties of the layers used to implement them. for example, the properties of ip, tcp, and tls apply to dns over tls implementations." 202 {42, 109, 15, 52, 86, 95} {15: 0, 42: 1, 109: 1, 52: 2, 86: 3, 95: 3} {202: 0, 203: 1, 201: 1, 200: 2} "A DoH implementation is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate different queries to the same identity. DNS transports will generally carry the same privacy properties of the layers used to implement them. For example, the properties of IP, TCP, and TLS apply to DNS over TLS implementations." this pr is against pr c61b08d0605e451d66ea5538d0e6b1ab13146b9f c61b08d0605e451d66ea5538d0e6b1ab13146b9f draft-ietf-doh-dns-over-https-latest 1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde ['1a59948f89eb7ded4c6ed7d77c6a86dc00a3087e2bf6bb3e7062d8cf9ccb5fde', '5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176'] B "a doh implementation is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate different queries to the same identity. dns transports will generally carry the same privacy properties of the layers used to implement them. for example, the properties of ip, tcp, and tls apply to dns over tls implementations." 202 {42, 109, 15, 52, 86, 95} {86: 0, 52: 1, 42: 2, 109: 2, 95: 2, 15: 3} {202: 0, 200: 1, 203: 2, 201: 2} "Determining whether or not a DoH implementation requires HTTP cookie RFC6265 support is particularly important because HTTP cookies are the primary state tracking mechanism in HTTP. HTTP Cookies SHOULD NOT be accepted by DOH clients unless they are explicitly required by a use case." this pr is against pr c61b08d0605e451d66ea5538d0e6b1ab13146b9f c61b08d0605e451d66ea5538d0e6b1ab13146b9f draft-ietf-doh-dns-over-https-latest beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b ['beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b'] B "determining whether or not a doh implementation requires http cookie rfc6265 support is particularly important because http cookies are the primary state tracking mechanism in http. http cookies should not be accepted by doh clients unless they are explicitly required by a use case." 201 {42, 109, 15, 52, 86, 95} {109: 0, 52: 1, 15: 1, 42: 2, 86: 2, 95: 2} {201: 0, 200: 1, 202: 1, 203: 2} "A DoH application is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate different queries to the same identity. DoH inherits the privacy properties of the HTTPS stack and is not known to introduce new concerns beyond that of HTTPS." This PR is against the privacy considerations branch ( pr ) 2205ef7b88461c5928ed59423af24a4a615c3c8e 2205ef7b88461c5928ed59423af24a4a615c3c8e draft-ietf-doh-dns-over-https-latest e747dc43c43382a4c4f5ddfee55ec6e24f2dc2de7cf176dfaeab85a9a99d4237 ['df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'e747dc43c43382a4c4f5ddfee55ec6e24f2dc2de7cf176dfaeab85a9a99d4237', '5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176'] A "a doh application is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate different queries to the same identity. doh inherits the privacy properties of the https stack and is not known to introduce new concerns beyond that of https." 201 {42, 109, 15, 52, 86, 95} {109: 0, 52: 1, 15: 1, 42: 2, 86: 2, 95: 2} {201: 0, 200: 1, 202: 1, 203: 2} "A DoH application is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate different queries to the same identity. DNS transports will generally carry the same privacy properties of the layers used to implement them. For example, the properties of IP, TCP, and TLS apply to DNS over TLS implementations." This PR is against the privacy considerations branch ( pr ) 2205ef7b88461c5928ed59423af24a4a615c3c8e 2205ef7b88461c5928ed59423af24a4a615c3c8e draft-ietf-doh-dns-over-https-latest 5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176 ['df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'e747dc43c43382a4c4f5ddfee55ec6e24f2dc2de7cf176dfaeab85a9a99d4237', '5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176'] B "a doh application is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate different queries to the same identity. dns transports will generally carry the same privacy properties of the layers used to implement them. for example, the properties of ip, tcp, and tls apply to dns over tls implementations." 201 {42, 109, 15, 52, 86, 95} {109: 0, 52: 1, 15: 1, 42: 2, 86: 2, 95: 2} {201: 0, 200: 1, 202: 1, 203: 2} "The privacy considerations of using the HTTPS layer in DoH are incremental to those of DNS over TLS. DoH is not known to introduce new concerns beyond those associated with HTTPS." This PR is against the privacy considerations branch ( pr ) 2205ef7b88461c5928ed59423af24a4a615c3c8e 2205ef7b88461c5928ed59423af24a4a615c3c8e draft-ietf-doh-dns-over-https-latest df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022 ['df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'e747dc43c43382a4c4f5ddfee55ec6e24f2dc2de7cf176dfaeab85a9a99d4237', '5b869a8530c4d77da3ff4fee44ee18ca316e9b3e1c22a5704a159347d0afd176'] B "the privacy considerations of using the https layer in doh are incremental to those of dns over tls. doh is not known to introduce new concerns beyond those associated with https." 201 {} {} {} "Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features, and their deployment context, when deciding whether or not to enable them." This PR is against the privacy considerations branch ( pr ) 2205ef7b88461c5928ed59423af24a4a615c3c8e 2205ef7b88461c5928ed59423af24a4a615c3c8e draft-ietf-doh-dns-over-https-latest 98fb1053401d4cd8107120655bb17c4f5a59f10647da06fb53a6f95c041a0721 ['98fb1053401d4cd8107120655bb17c4f5a59f10647da06fb53a6f95c041a0721', '40a643ad87c98aa335f4987990afa4797f6c5a2a9c1bf99850008c0bf2593237'] A "implementations of doh clients and servers need to consider the benefit and privacy impact of all these features, and their deployment context, when deciding whether or not to enable them." 201 {} {} {} "The DoH protocol design allows applications to fully leverage all the features of the HTTP ecosystem, including features not enumerated here. Implementations of DoH clients and servers need to consider the benefit and privacy impact of these features, and their deployment context, when deciding whether or not to enable them. Implementations should expose the minimal set of data needed to achieve the desired feature set." This PR is against the privacy considerations branch ( pr ) 2205ef7b88461c5928ed59423af24a4a615c3c8e 2205ef7b88461c5928ed59423af24a4a615c3c8e draft-ietf-doh-dns-over-https-latest 40a643ad87c98aa335f4987990afa4797f6c5a2a9c1bf99850008c0bf2593237 ['98fb1053401d4cd8107120655bb17c4f5a59f10647da06fb53a6f95c041a0721', '40a643ad87c98aa335f4987990afa4797f6c5a2a9c1bf99850008c0bf2593237'] B "the doh protocol design allows applications to fully leverage all the features of the http ecosystem, including features not enumerated here. implementations of doh clients and servers need to consider the benefit and privacy impact of these features, and their deployment context, when deciding whether or not to enable them. implementations should expose the minimal set of data needed to achieve the desired feature set." 200 {} {} {} "PrivacyConsiderations and Security discuss additional considerations for the integration with HTTP." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest 98dea4c04278c8ae69795fc468ca5a3de6f6c0a0717e86eb4ac52b3ce04dce50 ['98dea4c04278c8ae69795fc468ca5a3de6f6c0a0717e86eb4ac52b3ce04dce50'] B "privacyconsiderations and security discuss additional considerations for the integration with http." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "RFC7626 discusses DNS Privacy Considerations in both ""On the wire"" (Section 2.4), and ""In the server"" (Section 2.5) contexts. This is also a useful framing for DoH's privacy considerations." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "rfc7626 discusses dns privacy considerations in both ""on the wire"" (section 2.4), and ""in the server"" (section 2.5) contexts. this is also a useful framing for doh's privacy considerations." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "DoH encrypts DNS traffic and requires authentication of the server. This mitigates both passive surveillance RFC7258 and active attacks that attempt to divert DNS traffic to rogue servers (RFC7626 Section 2.5.1). DNS over TLS RFC7858 provides similar protections, while direct UDP and TCP based transports are vulnerable to this class of attack." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest 8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "doh encrypts dns traffic and requires authentication of the server. this mitigates both passive surveillance rfc7258 and active attacks that attempt to divert dns traffic to rogue servers (rfc7626 section 2.5.1). dns over tls rfc7858 provides similar protections, while direct udp and tcp based transports are vulnerable to this class of attack." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "Additionally, the use of the HTTPS default port 443 and the ability to mix DoH traffic with other HTTPS traffic on the same connection can deter unprivileged on-path devices from interfering with DNS operations and make DNS traffic analysis more difficult." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "additionally, the use of the https default port 443 and the ability to mix doh traffic with other https traffic on the same connection can deter unprivileged on-path devices from interfering with dns operations and make dns traffic analysis more difficult." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "A DoH implementation is built on IP, TCP, TLS, and HTTP. Each layer contains one or more common features that can be used to correlate queries to the same identity. DNS transports will generally carry the same privacy properties of the layers used to implement them. For example, the properties of IP, TCP, and TLS apply to DNS over TLS implementations." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest 8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "a doh implementation is built on ip, tcp, tls, and http. each layer contains one or more common features that can be used to correlate queries to the same identity. dns transports will generally carry the same privacy properties of the layers used to implement them. for example, the properties of ip, tcp, and tls apply to dns over tls implementations." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "The privacy considerations of using the HTTPS layer in DoH are incremental to those of DNS over TLS. DoH is not known to introduce new concerns beyond those associated with HTTPS." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "the privacy considerations of using the https layer in doh are incremental to those of dns over tls. doh is not known to introduce new concerns beyond those associated with https." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "At the IP level, the client address provides obvious correlation information. This can be mitigated by use of a NAT, proxy, VPN, or simple address rotation over time. It may be aggravated by use of a DNS server that can correlate real-time addressing information with other personal identifiers, such as when a DNS server and DHCP server are operated by the same entity." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest 0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "at the ip level, the client address provides obvious correlation information. this can be mitigated by use of a nat, proxy, vpn, or simple address rotation over time. it may be aggravated by use of a dns server that can correlate real-time addressing information with other personal identifiers, such as when a dns server and dhcp server are operated by the same entity." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "TCP-based solutions may seek performance through the use of TCP Fast Open RFC7413. The cookies used in TCP Fast Open allow servers to correlate TCP sessions together." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest 76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "tcp-based solutions may seek performance through the use of tcp fast open rfc7413. the cookies used in tcp fast open allow servers to correlate tcp sessions together." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "TLS based implementations often achieve better handshake performance through the use of some form of session resumption mechanism such as session tickets RFC5077. Session resumption creates trivial mechanisms for a server to correlate TLS connections together." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "tls based implementations often achieve better handshake performance through the use of some form of session resumption mechanism such as session tickets rfc5077. session resumption creates trivial mechanisms for a server to correlate tls connections together." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "HTTP's feature set can also be used for identification and tracking in a number of different ways. For example, authentication request header fields explicitly identify profiles in use, and HTTP Cookies are designed as an explicit state tracking mechanism between the client and serving site and often are used as an authentication mechanism." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "http's feature set can also be used for identification and tracking in a number of different ways. for example, authentication request header fields explicitly identify profiles in use, and http cookies are designed as an explicit state tracking mechanism between the client and serving site and often are used as an authentication mechanism." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "Additionally, the User-Agent and Accept-Language request header fields often convey specific information about the client version or locale. This facilitates content negotiation and operational work- arounds for implementation bugs. Request header fields that control caching can expose state information about a subset of the client's history. Mixing DoH requests with other HTTP requests on the same connection also provides an opportunity for richer data correlation." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "additionally, the user-agent and accept-language request header fields often convey specific information about the client version or locale. this facilitates content negotiation and operational work- arounds for implementation bugs. request header fields that control caching can expose state information about a subset of the client's history. mixing doh requests with other http requests on the same connection also provides an opportunity for richer data correlation." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "The DoH protocol design allows applications to fully leverage all the features of the HTTP ecosystem, including features not enumerated here. Implementations of DoH clients and servers need to consider the benefit and privacy impact of these features, and their deployment context, when deciding whether or not to enable them. Implementations are advised to expose the minimal set of data needed to achieve the desired feature set." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest d21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4 ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "the doh protocol design allows applications to fully leverage all the features of the http ecosystem, including features not enumerated here. implementations of doh clients and servers need to consider the benefit and privacy impact of these features, and their deployment context, when deciding whether or not to enable them. implementations are advised to expose the minimal set of data needed to achieve the desired feature set." 200 {42, 109, 15, 52, 86, 95} {52: 0, 42: 1, 109: 1, 86: 1, 95: 1, 15: 2} {200: 0, 203: 1, 201: 1, 202: 1} "Determining whether or not a DoH implementation requires HTTP cookie RFC6265 support is particularly important because HTTP cookies are the primary state tracking mechanism in HTTP. HTTP Cookies SHOULD NOT be accepted by DOH clients unless they are explicitly required by a use case." "Is there a reason not to make a recommendation for the case of a DOH - only service ? The current text says : Implementations of DoH clients and servers need to consider the benefit and privacy impact of all these features , and their deployment context , when deciding whether or not to enable them . Would you consider a recommendation like "" For DOH clients which do not intermingle DOH requests with other HTTP suppression of these headers and other potentially identifying headers is an appropriate data minimization strategy . "" ? That could be , but it would need a lot more text because your proposed text only deals with one layer ( HTTP ) while the text above it also talks about fingerprinting with TLS , TCP , and IP ( in reverse order ) . The WG can decide whether adding recommendations for just one layer is worthwhile or possibly misleading about the overall privacy . It was only an example , and I would be happy to expand . That said , I think this is the most likely source of leakage : folks using a general purpose HTTP subsystem that carries information not needed for DOH requests . Actually recommending data minimization there , rather than saying "" it should be considered "" seems like a worthwhile thing to do in my personal opinion . Agreed with ( on the list ) that this conversation should be on the list . Locking it here . I have unlocked the comments so that people can propose specific wording changes as diffs in the PR itself ; most questions and suggestions should be done on the list ." b72c9003e0b795373d848f157d1fb337b6f9d706 31c74913ebe5b1869c4f7bedfbff4fe4af9be9b1 draft-ietf-doh-dns-over-https-latest beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b ['76dadc18a159afcba0c68df11242f1ebb647c8a8f0fa85f60ed42bf6a58c02e3', 'aebf83bfe9a24a3fd53bf9e39359f6b0a43eaf8e88e176796fccb5b8e98ab684', 'df1baa22763e9b72ebe32b5365315c2d5831aebe6297416af1d38df008983022', 'c1e07e5e999b352661b6850a111da8cc15cd529e89e6124a0abfafee2266f44b', 'e5fdd3473397b9a4535e7231678bf75e4858679223409b79de8ec32f0c462cd7', '8a44899c800b24df7ef90fa04f7406f94edd43ca63823d4e3a2a46163b6e6f68', '0d4ff99dabbad7b8e5abd7489da237d4c55f142ee4168665e2ac3a854499fdff', 'af15a41c3222aeb97da89fe3eae978b246f0635bf9d43571b5e912f3bcd05bad', 'd21c2927eff87493e6f67672b0c013a056c128551abe48680688226ca18a6ec4', 'beaebe77c44a798f54b0ec8c9f4f6a28cb4d1d6fd4210470558c3a9254d67b6b', '8561013166604586ff1b84fb0710ce8d0e20dccfc31e90ce13ae788f01a59608', 'a23ecb77425f6236e8a6d5a56206a6da64dd86c6cb614660cb9ca444bb181979'] B "determining whether or not a doh implementation requires http cookie rfc6265 support is particularly important because http cookies are the primary state tracking mechanism in http. http cookies should not be accepted by doh clients unless they are explicitly required by a use case." 199 {} {} {} "A server that supports this protocol is called a ""DNS API server"" to differentiate it from a ""DNS server"" (one that only provides DNS service over one or more of the other transport protocols standardized for DNS). Similarly, a client that supports this protocol is called a ""DNS API client""." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 48bd3cc4eba8d2771bee2ddc335a52a5c1d27b2e02b661a341afa1d79856e1d6 ['282206453e6cb18d0fdf153de7c426099ff561227c733eaf7316ed8b1ee72df2', '48bd3cc4eba8d2771bee2ddc335a52a5c1d27b2e02b661a341afa1d79856e1d6'] A "a server that supports this protocol is called a ""dns api server"" to differentiate it from a ""dns server"" (one that only provides dns service over one or more of the other transport protocols standardized for dns). similarly, a client that supports this protocol is called a ""dns api client""." 199 {} {} {} "A server that supports this protocol is called a ""DoH server"" to differentiate it from a ""DNS server"" (one that only provides DNS service over one or more of the other transport protocols standardized for DNS). Similarly, a client that supports this protocol is called a ""DoH client""." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 282206453e6cb18d0fdf153de7c426099ff561227c733eaf7316ed8b1ee72df2 ['282206453e6cb18d0fdf153de7c426099ff561227c733eaf7316ed8b1ee72df2', '48bd3cc4eba8d2771bee2ddc335a52a5c1d27b2e02b661a341afa1d79856e1d6'] B "a server that supports this protocol is called a ""doh server"" to differentiate it from a ""dns server"" (one that only provides dns service over one or more of the other transport protocols standardized for dns). similarly, a client that supports this protocol is called a ""doh client""." 199 {} {} {} "Configuration, discovery, and updating of the URI Template RFC6570 (see httprequest) is done out of band from this protocol. Note that configuration might be manual (such as a user typing URI Templates in a user interface for ""options"") or automatic (such as URI Templates being supplied in responses from DHCP or similar protocols). DNS API Servers MAY support more than one URI. This allows the different endpoints to have different properties such as different authentication requirements or service level guarantees." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82 ['c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82', '02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d', '635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9', '5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71', '06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec', '2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031'] A "configuration, discovery, and updating of the uri template rfc6570 (see httprequest) is done out of band from this protocol. note that configuration might be manual (such as a user typing uri templates in a user interface for ""options"") or automatic (such as uri templates being supplied in responses from dhcp or similar protocols). dns api servers may support more than one uri. this allows the different endpoints to have different properties such as different authentication requirements or service level guarantees." 199 {} {} {} "A DNS API client uses configuration to select the URI, and thus the DNS API server, that is to be used for resolution. RFC2818 defines how HTTPS verifies the DNS API server's identity." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031 ['c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82', '02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d', '635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9', '5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71', '06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec', '2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031'] A "a dns api client uses configuration to select the uri, and thus the dns api server, that is to be used for resolution. rfc2818 defines how https verifies the dns api server's identity." 199 {} {} {} "A DNS API client MUST NOT use a different URI simply because it was discovered outside of the client's configuration, or because a server offers an unsolicited response that appears to be a valid answer to a DNS query. This specification does not extend DNS resolution privileges to URIs that are not recognized by the DNS API client as configured URIs. Such scenarios may create additional operational, tracking, and security hazards that require limitations for safe usage. A future specification may support this use case." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d ['c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82', '02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d', '635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9', '5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71', '06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec', '2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031'] A "a dns api client must not use a different uri simply because it was discovered outside of the client's configuration, or because a server offers an unsolicited response that appears to be a valid answer to a dns query. this specification does not extend dns resolution privileges to uris that are not recognized by the dns api client as configured uris. such scenarios may create additional operational, tracking, and security hazards that require limitations for safe usage. a future specification may support this use case." 199 {} {} {} "Configuration, discovery, and updating of the URI Template RFC6570 (see httprequest) is done out of band from this protocol. Note that configuration might be manual (such as a user typing URI Templates in a user interface for ""options"") or automatic (such as URI Templates being supplied in responses from DHCP or similar protocols). DoH Servers MAY support more than one URI. This allows the different endpoints to have different properties such as different authentication requirements or service level guarantees." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9 ['c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82', '02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d', '635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9', '5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71', '06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec', '2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031'] B "configuration, discovery, and updating of the uri template rfc6570 (see httprequest) is done out of band from this protocol. note that configuration might be manual (such as a user typing uri templates in a user interface for ""options"") or automatic (such as uri templates being supplied in responses from dhcp or similar protocols). doh servers may support more than one uri. this allows the different endpoints to have different properties such as different authentication requirements or service level guarantees." 199 {} {} {} "A DoH client uses configuration to select the URI, and thus the DoH server, that is to be used for resolution. RFC2818 defines how HTTPS verifies the DoH server's identity." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec ['c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82', '02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d', '635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9', '5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71', '06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec', '2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031'] B "a doh client uses configuration to select the uri, and thus the doh server, that is to be used for resolution. rfc2818 defines how https verifies the doh server's identity." 199 {} {} {} "A DoH client MUST NOT use a different URI simply because it was discovered outside of the client's configuration, or because a server offers an unsolicited response that appears to be a valid answer to a DNS query. This specification does not extend DNS resolution privileges to URIs that are not recognized by the DoH client as configured URIs. Such scenarios may create additional operational, tracking, and security hazards that require limitations for safe usage. A future specification may support this use case." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71 ['c1f44fd90839f6952d6ca17219310ccc22ca9c75e22356ea74b5cb999b43dc82', '02beac685937b33b70ef19c2eb10f2bbb31b040c43db050e38a14ee5c73e146d', '635ebfa8afad50fe9be697c20108712792e87949aa1c84a92525bb60ba1b89e9', '5ba61c8c3be2826fe5d686f23df428f7346db0b5381b34d3309f2d5354392e71', '06a78fd89991b1c91c6bb7808d4421c063fd1d8121b6c6014c819a26ec8011ec', '2bc808105da7b373c68a13e17a214276a1468c0ec8ece91d0010f493940bd031'] B "a doh client must not use a different uri simply because it was discovered outside of the client's configuration, or because a server offers an unsolicited response that appears to be a valid answer to a dns query. this specification does not extend dns resolution privileges to uris that are not recognized by the doh client as configured uris. such scenarios may create additional operational, tracking, and security hazards that require limitations for safe usage. a future specification may support this use case." 199 {} {} {} "A DNS API client encodes a single DNS query into an HTTP request using either the HTTP GET or POST method and the other requirements of this section. The DNS API server defines the URI used by the request through the use of a URI Template." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest a620f0b3e55ad318a036069b27c4d09248c65862c528af20af6658ed6c4299b7 ['a620f0b3e55ad318a036069b27c4d09248c65862c528af20af6658ed6c4299b7', '88977c40b92b1dd63d7c88a037f924d41365340af1b3bacc6b84e6c3ca3bcb8e'] A "a dns api client encodes a single dns query into an http request using either the http get or post method and the other requirements of this section. the dns api server defines the uri used by the request through the use of a uri template." 199 {} {} {} "A DoH client encodes a single DNS query into an HTTP request using either the HTTP GET or POST method and the other requirements of this section. The DoH server defines the URI used by the request through the use of a URI Template." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 88977c40b92b1dd63d7c88a037f924d41365340af1b3bacc6b84e6c3ca3bcb8e ['a620f0b3e55ad318a036069b27c4d09248c65862c528af20af6658ed6c4299b7', '88977c40b92b1dd63d7c88a037f924d41365340af1b3bacc6b84e6c3ca3bcb8e'] B "a doh client encodes a single dns query into an http request using either the http get or post method and the other requirements of this section. the doh server defines the uri used by the request through the use of a uri template." 199 {65, 67, 36, 73, 10, 75, 113} {113: 0, 73: 1, 65: 2, 67: 2, 36: 2, 10: 3, 75: 3} {199: 0, 131: 1, 61: 2, 133: 2, 128: 2, 6: 3, 33: 3} DNS API servers MUST implement both the POST and GET methods. "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc ['ef6cc621696436f9806f7fea23b4ab9826d08fd58f7206014e862923a6f608af', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] A dns api servers must implement both the post and get methods. 199 {65, 67, 36, 73, 10, 75, 113} {113: 0, 73: 1, 65: 2, 67: 2, 36: 2, 10: 3, 75: 3} {199: 0, 131: 1, 61: 2, 133: 2, 128: 2, 6: 3, 33: 3} DoH servers MUST implement both the POST and GET methods. "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest ef6cc621696436f9806f7fea23b4ab9826d08fd58f7206014e862923a6f608af ['ef6cc621696436f9806f7fea23b4ab9826d08fd58f7206014e862923a6f608af', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] B doh servers must implement both the post and get methods. 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "The DNS API client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""application/dns-message"" (as described in dnswire) responses but MAY also process any other type it receives." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4 ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] A "the dns api client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""application/dns-message"" (as described in dnswire) responses but may also process any other type it receives." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as application/dns-message, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as application/dns-message. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest d0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] A "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as application/dns-message, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as application/dns-message. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "DNS API clients can use HTTP/2 padding and compression in the same way that other HTTP/2 clients use (or don't use) them." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80 ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] A "dns api clients can use http/2 padding and compression in the same way that other http/2 clients use (or don't use) them." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "These examples use a DNS API service with a URI Template of ""https://dnsserver.example.net/dns-query{?dns}"" to resolve IN A records." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] A "these examples use a dns api service with a uri template of ""https://dnsserver.example.net/dns-query{?dns}"" to resolve in a records." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "The DoH client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""application/dns-message"" (as described in dnswire) responses but MAY also process any other type it receives." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] B "the doh client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""application/dns-message"" (as described in dnswire) responses but may also process any other type it receives." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "In order to maximize cache friendliness, DoH clients using media formats that include DNS ID, such as application/dns-message, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as application/dns-message. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] B "in order to maximize cache friendliness, doh clients using media formats that include dns id, such as application/dns-message, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as application/dns-message. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "DoH clients can use HTTP/2 padding and compression in the same way that other HTTP/2 clients use (or don't use) them." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] B "doh clients can use http/2 padding and compression in the same way that other http/2 clients use (or don't use) them." 199 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {53: 0, 107: 1, 54: 1, 106: 2, 5: 2, 64: 3, 4: 3, 12: 4, 98: 4, 99: 5} {199: 0, 144: 1, 131: 1, 142: 2, 65: 2, 61: 3, 78: 3, 20: 4} "These examples use a DoH service with a URI Template of ""https://dnsserver.example.net/dns-query{?dns}"" to resolve IN A records." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b ['78b738448f4c0f18f6160f12c665ec4fb8b3482129cbdedbc2a0769271ffc13d', 'be32709bf1f732bdce130c5adf3c121d7df881ff0d74cc5eee02c5b01772cf7b', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b', 'cbe307e03710493071d1aeffa45c710f72f30aaffb99591239ba4fdb9723780a', '7e7af7255ec1164caf13c54732af1618f4fa11b71d01a4e8d83380ce03eb8f80', '97d0f96ef743e9829bc932c95d0a8f2f0e20315e52ad331bb61a4a0899069e8b', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] B "these examples use a doh service with a uri template of ""https://dnsserver.example.net/dns-query{?dns}"" to resolve in a records." 199 {} {} {} "The only response type defined in this document is ""application/dns- message"", but it is possible that other response formats will be defined in the future. A DNS API server MUST be able to process application/dns-message request messages." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 7563d1cb651c18fe47dbcdf20d0e5053ab3e7f43ee1fd3608fa060769a5361b5 ['805c5cac7379c0ad02be5c066e08630bfe8eac5c10a51e265ba374249a1a4447', '7563d1cb651c18fe47dbcdf20d0e5053ab3e7f43ee1fd3608fa060769a5361b5'] A "the only response type defined in this document is ""application/dns- message"", but it is possible that other response formats will be defined in the future. a dns api server must be able to process application/dns-message request messages." 199 {} {} {} "The only response type defined in this document is ""application/dns- message"", but it is possible that other response formats will be defined in the future. A DoH server MUST be able to process application/dns-message request messages." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 805c5cac7379c0ad02be5c066e08630bfe8eac5c10a51e265ba374249a1a4447 ['805c5cac7379c0ad02be5c066e08630bfe8eac5c10a51e265ba374249a1a4447', '7563d1cb651c18fe47dbcdf20d0e5053ab3e7f43ee1fd3608fa060769a5361b5'] B "the only response type defined in this document is ""application/dns- message"", but it is possible that other response formats will be defined in the future. a doh server must be able to process application/dns-message request messages." 199 {} {} {} "HTTP responses with non-successful HTTP status codes do not contain replies to the original DNS question in the HTTP request. DNS API clients need to use the same semantic processing of non-successful HTTP status codes as other HTTP clients. This might mean that the DNS API client retries the query with the same DNS API server, such as authorization failures (HTTP status code 401 RFC7235 Section 3.1). It could also mean that the DNS API client retries with a different DNS API server, such as for unsupported media types (HTTP status code 415, RFC7231 Section 6.5.13), or where the server cannot generate a representation suitable for the client (HTTP status code 406, RFC7231 Section 6.5.6), and so on." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest ccba1b2281556f4fe7fa421e3d9b0d49f930cf4b4e67238c2f85ee644d2808ad ['5d093fba0515a931018b5ae48e760b773d75ba1bd2c504c24afa5419d5a40b58', 'ccba1b2281556f4fe7fa421e3d9b0d49f930cf4b4e67238c2f85ee644d2808ad'] A "http responses with non-successful http status codes do not contain replies to the original dns question in the http request. dns api clients need to use the same semantic processing of non-successful http status codes as other http clients. this might mean that the dns api client retries the query with the same dns api server, such as authorization failures (http status code 401 rfc7235 section 3.1). it could also mean that the dns api client retries with a different dns api server, such as for unsupported media types (http status code 415, rfc7231 section 6.5.13), or where the server cannot generate a representation suitable for the client (http status code 406, rfc7231 section 6.5.6), and so on." 199 {} {} {} "HTTP responses with non-successful HTTP status codes do not contain replies to the original DNS question in the HTTP request. DoH clients need to use the same semantic processing of non-successful HTTP status codes as other HTTP clients. This might mean that the DoH client retries the query with the same DoH server, such as authorization failures (HTTP status code 401 RFC7235 Section 3.1). It could also mean that the DoH client retries with a different DoH server, such as for unsupported media types (HTTP status code 415, RFC7231 Section 6.5.13), or where the server cannot generate a representation suitable for the client (HTTP status code 406, RFC7231 Section 6.5.6), and so on." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 5d093fba0515a931018b5ae48e760b773d75ba1bd2c504c24afa5419d5a40b58 ['5d093fba0515a931018b5ae48e760b773d75ba1bd2c504c24afa5419d5a40b58', 'ccba1b2281556f4fe7fa421e3d9b0d49f930cf4b4e67238c2f85ee644d2808ad'] B "http responses with non-successful http status codes do not contain replies to the original dns question in the http request. doh clients need to use the same semantic processing of non-successful http status codes as other http clients. this might mean that the doh client retries the query with the same doh server, such as authorization failures (http status code 401 rfc7235 section 3.1). it could also mean that the doh client retries with a different doh server, such as for unsupported media types (http status code 415, rfc7231 section 6.5.13), or where the server cannot generate a representation suitable for the client (http status code 406, rfc7231 section 6.5.6), and so on." 199 {0, 1, 74, 43, 62} {74: 0, 1: 1, 0: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "A DOH exchange can pass through a hierarchy of caches that include both HTTP and DNS specific caches. These caches may exist beteen the DNS API server and client, or on the DNS API client itself. HTTP caches are by design generic; that is, they do not understand this protocol. Even if a DNS API client has modified its cache implementation to be aware of DOH semantics, it does not follow that all upstream caches (for example, inline proxies, server-side gateways and Content Delivery Networks) will be." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e ['9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e', '7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', '2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf'] A "a doh exchange can pass through a hierarchy of caches that include both http and dns specific caches. these caches may exist beteen the dns api server and client, or on the dns api client itself. http caches are by design generic; that is, they do not understand this protocol. even if a dns api client has modified its cache implementation to be aware of doh semantics, it does not follow that all upstream caches (for example, inline proxies, server-side gateways and content delivery networks) will be." 199 {0, 1, 74, 43, 62} {74: 0, 1: 1, 0: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "As a result, DNS API servers need to carefully consider the HTTP caching metadata they send in response to GET requests (POST requests are not cacheable unless specific response headers are sent; this is not widely implemented, and not advised for DOH)." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4 ['9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e', '7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', '2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf'] A "as a result, dns api servers need to carefully consider the http caching metadata they send in response to get requests (post requests are not cacheable unless specific response headers are sent; this is not widely implemented, and not advised for doh)." 199 {0, 1, 74, 43, 62} {74: 0, 1: 1, 0: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "In particular, DNS API servers SHOULD assign an explicit freshness lifetime (RFC7234 Section 4.2) so that the DNS API client is more likely to use fresh DNS data. This requirement is due to HTTP caches being able to assign their own heuristic freshness (such as that described in RFC7234 Section 4.2.2), which would take control of the cache contents out of the hands of the DNS API server." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a ['9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e', '7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', '2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf'] A "in particular, dns api servers should assign an explicit freshness lifetime (rfc7234 section 4.2) so that the dns api client is more likely to use fresh dns data. this requirement is due to http caches being able to assign their own heuristic freshness (such as that described in rfc7234 section 4.2.2), which would take control of the cache contents out of the hands of the dns api server." 199 {0, 1, 74, 43, 62} {74: 0, 1: 1, 0: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "A DoH exchange can pass through a hierarchy of caches that include both HTTP and DNS specific caches. These caches may exist beteen the DoH server and client, or on the DoH client itself. HTTP caches are by design generic; that is, they do not understand this protocol. Even if a DoH client has modified its cache implementation to be aware of DoH semantics, it does not follow that all upstream caches (for example, inline proxies, server-side gateways and Content Delivery Networks) will be." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04 ['9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e', '7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', '2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf'] B "a doh exchange can pass through a hierarchy of caches that include both http and dns specific caches. these caches may exist beteen the doh server and client, or on the doh client itself. http caches are by design generic; that is, they do not understand this protocol. even if a doh client has modified its cache implementation to be aware of doh semantics, it does not follow that all upstream caches (for example, inline proxies, server-side gateways and content delivery networks) will be." 199 {0, 1, 74, 43, 62} {74: 0, 1: 1, 0: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "As a result, DoH servers need to carefully consider the HTTP caching metadata they send in response to GET requests (POST requests are not cacheable unless specific response headers are sent; this is not widely implemented, and not advised for DoH)." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf ['9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e', '7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', '2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf'] B "as a result, doh servers need to carefully consider the http caching metadata they send in response to get requests (post requests are not cacheable unless specific response headers are sent; this is not widely implemented, and not advised for doh)." 199 {0, 1, 74, 43, 62} {74: 0, 1: 1, 0: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "In particular, DoH servers SHOULD assign an explicit freshness lifetime (RFC7234 Section 4.2) so that the DoH client is more likely to use fresh DNS data. This requirement is due to HTTP caches being able to assign their own heuristic freshness (such as that described in RFC7234 Section 4.2.2), which would take control of the cache contents out of the hands of the DoH server." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d ['9a3f802df8bb6dad5bc1133606ea45c274153f6a07f2ea262f38ac6b4811349d', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e', '7d2891ce017973cab19c316c6e1d8dccf784dff5e8e282cf3646c54eaeed1f04', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', '2390d5b47a0a6cdf8094c5bca7fee1e80c677830bddf9456dcb92777ce98abdf'] B "in particular, doh servers should assign an explicit freshness lifetime (rfc7234 section 4.2) so that the doh client is more likely to use fresh dns data. this requirement is due to http caches being able to assign their own heuristic freshness (such as that described in rfc7234 section 4.2.2), which would take control of the cache contents out of the hands of the doh server." 199 {0, 1, 74, 43, 62} {0: 0, 1: 1, 74: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "DNS API servers also need to consider caching when generating responses that are not globally valid. For instance, if a DNS API server customizes a response based on the client's identity, it would not want to allow global reuse of that response. This could be accomplished through a variety of HTTP techniques such as a Cache- Control max-age of 0, or by using the Vary response header (RFC7231 Section 7.1.4) to establish a secondary cache key (RFC7234 Section 4.1)." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', '57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8', '9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9'] A "dns api servers also need to consider caching when generating responses that are not globally valid. for instance, if a dns api server customizes a response based on the client's identity, it would not want to allow global reuse of that response. this could be accomplished through a variety of http techniques such as a cache- control max-age of 0, or by using the vary response header (rfc7231 section 7.1.4) to establish a secondary cache key (rfc7234 section 4.1)." 199 {0, 1, 74, 43, 62} {0: 0, 1: 1, 74: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "DNS API clients MUST account for the Age response header's value (RFC7234) when calculating the DNS TTL of a response. For example, if a RRset is received with a DNS TTL of 600, but the Age header indicates that the response has been cached for 250 seconds, the remaining lifetime of the RRset is 350 seconds." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', '57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8', '9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9'] A "dns api clients must account for the age response header's value (rfc7234) when calculating the dns ttl of a response. for example, if a rrset is received with a dns ttl of 600, but the age header indicates that the response has been cached for 250 seconds, the remaining lifetime of the rrset is 350 seconds." 199 {0, 1, 74, 43, 62} {0: 0, 1: 1, 74: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "DNS API clients can request an uncached copy of a response by using the ""no-cache"" request cache control directive (RFC7234, Section 5.2.1.4) and similar controls. Note that some caches might not honor these directives, either due to configuration or interaction with traditional DNS caches that do not have such a mechanism." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', '57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8', '9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9'] A "dns api clients can request an uncached copy of a response by using the ""no-cache"" request cache control directive (rfc7234, section 5.2.1.4) and similar controls. note that some caches might not honor these directives, either due to configuration or interaction with traditional dns caches that do not have such a mechanism." 199 {0, 1, 74, 43, 62} {0: 0, 1: 1, 74: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "DoH servers also need to consider caching when generating responses that are not globally valid. For instance, if a DoH server customizes a response based on the client's identity, it would not want to allow global reuse of that response. This could be accomplished through a variety of HTTP techniques such as a Cache- Control max-age of 0, or by using the Vary response header (RFC7231 Section 7.1.4) to establish a secondary cache key (RFC7234 Section 4.1)." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', '57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8', '9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9'] B "doh servers also need to consider caching when generating responses that are not globally valid. for instance, if a doh server customizes a response based on the client's identity, it would not want to allow global reuse of that response. this could be accomplished through a variety of http techniques such as a cache- control max-age of 0, or by using the vary response header (rfc7231 section 7.1.4) to establish a secondary cache key (rfc7234 section 4.1)." 199 {0, 1, 74, 43, 62} {0: 0, 1: 1, 74: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "DoH clients MUST account for the Age response header's value (RFC7234) when calculating the DNS TTL of a response. For example, if a RRset is received with a DNS TTL of 600, but the Age header indicates that the response has been cached for 250 seconds, the remaining lifetime of the RRset is 350 seconds." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', '57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8', '9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9'] B "doh clients must account for the age response header's value (rfc7234) when calculating the dns ttl of a response. for example, if a rrset is received with a dns ttl of 600, but the age header indicates that the response has been cached for 250 seconds, the remaining lifetime of the rrset is 350 seconds." 199 {0, 1, 74, 43, 62} {0: 0, 1: 1, 74: 2, 43: 2, 62: 2} {199: 0, 177: 1, 56: 2, 154: 2} "DoH clients can request an uncached copy of a response by using the ""no-cache"" request cache control directive (RFC7234, Section 5.2.1.4) and similar controls. Note that some caches might not honor these directives, either due to configuration or interaction with traditional DNS caches that do not have such a mechanism." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '674aaa5ce00b3326ff5707799a81e6653cf48ca1a75b90a9b99430a27c17d4ea', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', '57b399f00ef61d09aabe3d952153215d6c4ff9917e5a0e0f806521f924db17f8', '9770ffabe53b821ec054bcb471cbb2e6fc6c94609b1bf97ef43c0e775047a386', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9'] B "doh clients can request an uncached copy of a response by using the ""no-cache"" request cache control directive (rfc7234, section 5.2.1.4) and similar controls. note that some caches might not honor these directives, either due to configuration or interaction with traditional dns caches that do not have such a mechanism." 199 {32, 101, 102, 103, 40, 44, 45, 31} {103: 0, 40: 1, 45: 1, 32: 2, 102: 2, 31: 3, 101: 3, 44: 4} {199: 0, 174: 1, 169: 1, 157: 2, 144: 2, 63: 3, 142: 3, 128: 4} "Before using DOH response data for DNS resolution, the client MUST establish that the HTTP request URI may be used for the DOH query. For HTTP requests initiated by the DNS API client this is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest d6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103 ['d2bb9ccbdbc60226f665e3c7bc808ab040b16e91cb72935e33b36f0733d1512b', '31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221', 'd6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103', '391bb3ecf9a63bf20382af8eae1b8f4db67acf2d763ad55081463ed796246964'] A "before using doh response data for dns resolution, the client must establish that the http request uri may be used for the doh query. for http requests initiated by the dns api client this is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request." 199 {32, 101, 102, 103, 40, 44, 45, 31} {103: 0, 40: 1, 45: 1, 32: 2, 102: 2, 31: 3, 101: 3, 44: 4} {199: 0, 174: 1, 169: 1, 157: 2, 144: 2, 63: 3, 142: 3, 128: 4} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""application/dns-message"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4). Those media types MUST be flexible enough to express every DNS query that would normally be sent in DNS over UDP (including queries and responses that use DNS extensions, but not those that require multiple responses)." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221 ['d2bb9ccbdbc60226f665e3c7bc808ab040b16e91cb72935e33b36f0733d1512b', '31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221', 'd6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103', '391bb3ecf9a63bf20382af8eae1b8f4db67acf2d763ad55081463ed796246964'] A "in order to maximize interoperability, dns api clients and dns api servers must support the ""application/dns-message"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4). those media types must be flexible enough to express every dns query that would normally be sent in dns over udp (including queries and responses that use dns extensions, but not those that require multiple responses)." 199 {32, 101, 102, 103, 40, 44, 45, 31} {103: 0, 40: 1, 45: 1, 32: 2, 102: 2, 31: 3, 101: 3, 44: 4} {199: 0, 174: 1, 169: 1, 157: 2, 144: 2, 63: 3, 142: 3, 128: 4} "Before using DoH response data for DNS resolution, the client MUST establish that the HTTP request URI may be used for the DoH query. For HTTP requests initiated by the DoH client this is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest d2bb9ccbdbc60226f665e3c7bc808ab040b16e91cb72935e33b36f0733d1512b ['d2bb9ccbdbc60226f665e3c7bc808ab040b16e91cb72935e33b36f0733d1512b', '31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221', 'd6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103', '391bb3ecf9a63bf20382af8eae1b8f4db67acf2d763ad55081463ed796246964'] B "before using doh response data for dns resolution, the client must establish that the http request uri may be used for the doh query. for http requests initiated by the doh client this is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request." 199 {32, 101, 102, 103, 40, 44, 45, 31} {103: 0, 40: 1, 45: 1, 32: 2, 102: 2, 31: 3, 101: 3, 44: 4} {199: 0, 174: 1, 169: 1, 157: 2, 144: 2, 63: 3, 142: 3, 128: 4} "In order to maximize interoperability, DoH clients and DoH servers MUST support the ""application/dns-message"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4). Those media types MUST be flexible enough to express every DNS query that would normally be sent in DNS over UDP (including queries and responses that use DNS extensions, but not those that require multiple responses)." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 391bb3ecf9a63bf20382af8eae1b8f4db67acf2d763ad55081463ed796246964 ['d2bb9ccbdbc60226f665e3c7bc808ab040b16e91cb72935e33b36f0733d1512b', '31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221', 'd6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103', '391bb3ecf9a63bf20382af8eae1b8f4db67acf2d763ad55081463ed796246964'] B "in order to maximize interoperability, doh clients and doh servers must support the ""application/dns-message"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4). those media types must be flexible enough to express every dns query that would normally be sent in dns over udp (including queries and responses that use dns extensions, but not those that require multiple responses)." 199 {71, 72, 76, 77, 48, 55} {55: 0, 76: 1, 48: 2, 72: 2, 77: 2, 71: 3} {199: 0, 196: 1, 131: 2, 144: 2, 122: 2, 142: 3} "DNS API clients using this media type MAY have one or more EDNS options RFC6891 in the request. DNS API servers using this media type MUST ignore the value given for the EDNS UDP payload size in DNS requests." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c ['7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', 'fc1b39d098bb151656a4a7de97e58cf37267fc699c9a85fb9ff4935fe8a88001'] A "dns api clients using this media type may have one or more edns options rfc6891 in the request. dns api servers using this media type must ignore the value given for the edns udp payload size in dns requests." 199 {71, 72, 76, 77, 48, 55} {55: 0, 76: 1, 48: 2, 72: 2, 77: 2, 71: 3} {199: 0, 196: 1, 131: 2, 144: 2, 122: 2, 142: 3} "DoH clients using this media type MAY have one or more EDNS options RFC6891 in the request. DoH servers using this media type MUST ignore the value given for the EDNS UDP payload size in DNS requests." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest fc1b39d098bb151656a4a7de97e58cf37267fc699c9a85fb9ff4935fe8a88001 ['7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', 'fc1b39d098bb151656a4a7de97e58cf37267fc699c9a85fb9ff4935fe8a88001'] B "doh clients using this media type may have one or more edns options rfc6891 in the request. doh servers using this media type must ignore the value given for the edns udp payload size in dns requests." 199 {119, 87} {87: 0, 119: 1} {199: 0, 176: 1} "Session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (RFC7540 Section 10.6) and padding (RFC7540 Section 10.7 ). DNS API Servers can also add DNS padding RFC7830 if the DNS API requests it in the DNS query." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 6d35cac2472c7d8b3c9cfb074b7b690d1741a086b6724a5682c95bbd566162b1 ['bfe59ac1150c9fb78d8d56b50c927a385604b6a3a6301be7657d6288a61ec7b6', '0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2', 'afc98d2fd87b45705de9951aa84682eea6d90d55f803b23f9d4a0a95ac7c833a', '6d35cac2472c7d8b3c9cfb074b7b690d1741a086b6724a5682c95bbd566162b1'] A "session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (rfc7540 section 10.6) and padding (rfc7540 section 10.7 ). dns api servers can also add dns padding rfc7830 if the dns api requests it in the dns query." 199 {119, 87} {87: 0, 119: 1} {199: 0, 176: 1} "The HTTPS connection provides transport security for the interaction between the DNS API server and client, but does not provide the response integrity of DNS data provided by DNSSEC. DNSSEC and DOH are independent and fully compatible protocols, each solving different problems. The use of one does not diminish the need nor the usefulness of the other. It is the choice of a client to either perform full DNSSEC validation of answers or to trust the DNS API server to do DNSSEC validation and inspect the AD (Authentic Data) bit in the returned message to determine whether an answer was authentic or not. As noted in http-response, different response media types will provide more or less information from a DNS response so this choice may be affected by the response media type." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2 ['bfe59ac1150c9fb78d8d56b50c927a385604b6a3a6301be7657d6288a61ec7b6', '0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2', 'afc98d2fd87b45705de9951aa84682eea6d90d55f803b23f9d4a0a95ac7c833a', '6d35cac2472c7d8b3c9cfb074b7b690d1741a086b6724a5682c95bbd566162b1'] A "the https connection provides transport security for the interaction between the dns api server and client, but does not provide the response integrity of dns data provided by dnssec. dnssec and doh are independent and fully compatible protocols, each solving different problems. the use of one does not diminish the need nor the usefulness of the other. it is the choice of a client to either perform full dnssec validation of answers or to trust the dns api server to do dnssec validation and inspect the ad (authentic data) bit in the returned message to determine whether an answer was authentic or not. as noted in http-response, different response media types will provide more or less information from a dns response so this choice may be affected by the response media type." 199 {119, 87} {87: 0, 119: 1} {199: 0, 176: 1} "Session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (RFC7540 Section 10.6) and padding (RFC7540 Section 10.7 ). DoH Servers can also add DNS padding RFC7830 if the DoH requests it in the DNS query." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest bfe59ac1150c9fb78d8d56b50c927a385604b6a3a6301be7657d6288a61ec7b6 ['bfe59ac1150c9fb78d8d56b50c927a385604b6a3a6301be7657d6288a61ec7b6', '0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2', 'afc98d2fd87b45705de9951aa84682eea6d90d55f803b23f9d4a0a95ac7c833a', '6d35cac2472c7d8b3c9cfb074b7b690d1741a086b6724a5682c95bbd566162b1'] B "session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (rfc7540 section 10.6) and padding (rfc7540 section 10.7 ). doh servers can also add dns padding rfc7830 if the doh requests it in the dns query." 199 {119, 87} {87: 0, 119: 1} {199: 0, 176: 1} "The HTTPS connection provides transport security for the interaction between the DoH server and client, but does not provide the response integrity of DNS data provided by DNSSEC. DNSSEC and DoH are independent and fully compatible protocols, each solving different problems. The use of one does not diminish the need nor the usefulness of the other. It is the choice of a client to either perform full DNSSEC validation of answers or to trust the DoH server to do DNSSEC validation and inspect the AD (Authentic Data) bit in the returned message to determine whether an answer was authentic or not. As noted in http-response, different response media types will provide more or less information from a DNS response so this choice may be affected by the response media type." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest afc98d2fd87b45705de9951aa84682eea6d90d55f803b23f9d4a0a95ac7c833a ['bfe59ac1150c9fb78d8d56b50c927a385604b6a3a6301be7657d6288a61ec7b6', '0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2', 'afc98d2fd87b45705de9951aa84682eea6d90d55f803b23f9d4a0a95ac7c833a', '6d35cac2472c7d8b3c9cfb074b7b690d1741a086b6724a5682c95bbd566162b1'] B "the https connection provides transport security for the interaction between the doh server and client, but does not provide the response integrity of dns data provided by dnssec. dnssec and doh are independent and fully compatible protocols, each solving different problems. the use of one does not diminish the need nor the usefulness of the other. it is the choice of a client to either perform full dnssec validation of answers or to trust the doh server to do dnssec validation and inspect the ad (authentic data) bit in the returned message to determine whether an answer was authentic or not. as noted in http-response, different response media types will provide more or less information from a dns response so this choice may be affected by the response media type." 199 {} {} {} "In the absence of DNSSEC information, a DNS API server can give a client invalid data in response to a DNS query. selection disallows the use of DOH DNS responses that do not originate from configured servers. This prohibition does not guarantee protection against invalid data, but it does reduce the risk." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest be611ac2eb344cbf4c351a29f07c3d31b19c9732713c1dc33345e4ccedf8c798 ['906540333048e9d27a20b9ea3bcadec8a797abd33f55522cdbe3a609fc97bca4', 'be611ac2eb344cbf4c351a29f07c3d31b19c9732713c1dc33345e4ccedf8c798'] A "in the absence of dnssec information, a dns api server can give a client invalid data in response to a dns query. selection disallows the use of doh dns responses that do not originate from configured servers. this prohibition does not guarantee protection against invalid data, but it does reduce the risk." 199 {} {} {} "In the absence of DNSSEC information, a DoH server can give a client invalid data in response to a DNS query. selection disallows the use of DoH DNS responses that do not originate from configured servers. This prohibition does not guarantee protection against invalid data, but it does reduce the risk." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 906540333048e9d27a20b9ea3bcadec8a797abd33f55522cdbe3a609fc97bca4 ['906540333048e9d27a20b9ea3bcadec8a797abd33f55522cdbe3a609fc97bca4', 'be611ac2eb344cbf4c351a29f07c3d31b19c9732713c1dc33345e4ccedf8c798'] B "in the absence of dnssec information, a doh server can give a client invalid data in response to a dns query. selection disallows the use of doh dns responses that do not originate from configured servers. this prohibition does not guarantee protection against invalid data, but it does reduce the risk." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "The HTTPS channel used by this specification establishes secure two party communication between the DNS API client and the DNS API server. Filtering or inspection systems that rely on unsecured transport of DNS will not function in a DNS over HTTPS environment." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414 ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] A "the https channel used by this specification establishes secure two party communication between the dns api client and the dns api server. filtering or inspection systems that rely on unsecured transport of dns will not function in a dns over https environment." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "Some HTTPS client implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DNS API server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of OCSP RFC6960 servers or AIA for CRL fetching (RFC5280 Section 4.2.2.1) are examples of how this deadlock can happen. To mitigate the possibility of deadlock, DNS API servers SHOULD NOT rely on DNS based references to external resources in the TLS handshake. For OCSP the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC6066 Section 8 for TLS version 1.2. AIA deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. Note that these deadlocks also need to be considered for server that a DNS API server might redirect to." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813 ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] A "some https client implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the dns api server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of ocsp rfc6960 servers or aia for crl fetching (rfc5280 section 4.2.2.1) are examples of how this deadlock can happen. to mitigate the possibility of deadlock, dns api servers should not rely on dns based references to external resources in the tls handshake. for ocsp the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc6066 section 8 for tls version 1.2. aia deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. note that these deadlocks also need to be considered for server that a dns api server might redirect to." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "A DNS API client may face a similar bootstrapping problem when the HTTP request needs to resolve the hostname portion of the DNS URI. Just as the address of a traditional DNS nameserver cannot be originally determined from that same server, a DNS API client cannot use its DNS API server to initially resolve the server's host name into an address. Alternative strategies a client might employ include making the initial resolution part of the configuration, IP based URIs and corresponding IP based certificates for HTTPS, or resolving the DNS API server's hostname via traditional DNS or another DNS API server while still authenticating the resulting connection via HTTPS." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9 ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] A "a dns api client may face a similar bootstrapping problem when the http request needs to resolve the hostname portion of the dns uri. just as the address of a traditional dns nameserver cannot be originally determined from that same server, a dns api client cannot use its dns api server to initially resolve the server's host name into an address. alternative strategies a client might employ include making the initial resolution part of the configuration, ip based uris and corresponding ip based certificates for https, or resolving the dns api server's hostname via traditional dns or another dns api server while still authenticating the resulting connection via https." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "A DNS API server is allowed to answer queries with any valid DNS response. For example, a valid DNS response might have the TC (truncation) bit set in the DNS header to indicate that the server was not able to retrieve a full answer for the query but is providing the best answer it could get. A DNS API server can reply to queries with an HTTP error for queries that it cannot fulfill. In this same example, a DNS API server could use an HTTP error instead of a non- error response that has the TC bit set." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] A "a dns api server is allowed to answer queries with any valid dns response. for example, a valid dns response might have the tc (truncation) bit set in the dns header to indicate that the server was not able to retrieve a full answer for the query but is providing the best answer it could get. a dns api server can reply to queries with an http error for queries that it cannot fulfill. in this same example, a dns api server could use an http error instead of a non- error response that has the tc bit set." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "The HTTPS channel used by this specification establishes secure two party communication between the DoH client and the DoH server. Filtering or inspection systems that rely on unsecured transport of DNS will not function in a DNS over HTTPS environment." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062 ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] B "the https channel used by this specification establishes secure two party communication between the doh client and the doh server. filtering or inspection systems that rely on unsecured transport of dns will not function in a dns over https environment." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "Some HTTPS client implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DoH server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of OCSP RFC6960 servers or AIA for CRL fetching (RFC5280 Section 4.2.2.1) are examples of how this deadlock can happen. To mitigate the possibility of deadlock, DoH servers SHOULD NOT rely on DNS based references to external resources in the TLS handshake. For OCSP the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC6066 Section 8 for TLS version 1.2. AIA deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. Note that these deadlocks also need to be considered for server that a DoH server might redirect to." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] B "some https client implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the doh server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of ocsp rfc6960 servers or aia for crl fetching (rfc5280 section 4.2.2.1) are examples of how this deadlock can happen. to mitigate the possibility of deadlock, doh servers should not rely on dns based references to external resources in the tls handshake. for ocsp the server can bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc6066 section 8 for tls version 1.2. aia deadlocks can be avoided by providing intermediate certificates that might otherwise be obtained through additional requests. note that these deadlocks also need to be considered for server that a doh server might redirect to." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "A DoH client may face a similar bootstrapping problem when the HTTP request needs to resolve the hostname portion of the DNS URI. Just as the address of a traditional DNS nameserver cannot be originally determined from that same server, a DoH client cannot use its DoH server to initially resolve the server's host name into an address. Alternative strategies a client might employ include making the initial resolution part of the configuration, IP based URIs and corresponding IP based certificates for HTTPS, or resolving the DNS API server's hostname via traditional DNS or another DoH server while still authenticating the resulting connection via HTTPS." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest 7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] B "a doh client may face a similar bootstrapping problem when the http request needs to resolve the hostname portion of the dns uri. just as the address of a traditional dns nameserver cannot be originally determined from that same server, a doh client cannot use its doh server to initially resolve the server's host name into an address. alternative strategies a client might employ include making the initial resolution part of the configuration, ip based uris and corresponding ip based certificates for https, or resolving the dns api server's hostname via traditional dns or another doh server while still authenticating the resulting connection via https." 199 {111, 112, 21, 122, 124} {122: 0, 124: 1, 21: 1, 112: 2, 111: 3} {199: 0, 170: 1, 114: 1, 123: 2, 120: 3} "A DoH server is allowed to answer queries with any valid DNS response. For example, a valid DNS response might have the TC (truncation) bit set in the DNS header to indicate that the server was not able to retrieve a full answer for the query but is providing the best answer it could get. A DoH server can reply to queries with an HTTP error for queries that it cannot fulfill. In this same example, a DoH server could use an HTTP error instead of a non-error response that has the TC bit set." "This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" terminology : "" DNS API "" "" Fixed by" 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 0dc7ce3222c2f8a1f960ec4593d0c073596fc116 draft-ietf-doh-dns-over-https-latest fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4 ['fbd046d2ddc74395145b5b2cc7c7d06c3f8424d4e2a65514b78da86a1eba5ce4', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a', '27388f4f5a58cbd105b01ac25eb8b60ae2903bcf9de9ac99b75fdec9e3391414', '7f9217fc1c22c32aac174d8e975308f24ab2c92876c431a1cc3899288612781e', '2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '039118af14da1415ce16f980bbad68cd092a841c486760fe982dd37fbfbbf813', '8f4198a1d9e595ec1d7e1117cfadf439ad54a0a9e7ea861a8bb0182a822dd062', '676f11675e723678d368f80075eea678a44ed0ef859f83999351cb0a7a7e565a'] B "a doh server is allowed to answer queries with any valid dns response. for example, a valid dns response might have the tc (truncation) bit set in the dns header to indicate that the server was not able to retrieve a full answer for the query but is providing the best answer it could get. a doh server can reply to queries with an http error for queries that it cannot fulfill. in this same example, a doh server could use an http error instead of a non-error response that has the tc bit set." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "The data payload is the DNS on-the-wire format defined in RFC1035. The format is for DNS over UDP. Note that this is different than the wire format used in RFC7858. Also note that while RFC1035 says ""Messages carried by UDP are restricted to 512 bytes"", that was later updated by RFC6891. This protocol allows DNS on-the-wire format payloads of any size." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] A "the data payload is the dns on-the-wire format defined in rfc1035. the format is for dns over udp. note that this is different than the wire format used in rfc7858. also note that while rfc1035 says ""messages carried by udp are restricted to 512 bytes"", that was later updated by rfc6891. this protocol allows dns on-the-wire format payloads of any size." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "When using the GET method, the data payload MUST be encoded with base64url RFC4648 and then provided as a variable named ""dns"" to the URI Template expansion. Padding characters for base64url MUST NOT be included." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest 3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] A "when using the get method, the data payload must be encoded with base64url rfc4648 and then provided as a variable named ""dns"" to the uri template expansion. padding characters for base64url must not be included." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "When using the POST method, the data payload MUST NOT be encoded and is used directly as the HTTP message body." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] A "when using the post method, the data payload must not be encoded and is used directly as the http message body." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "DNS API clients using the DNS wire format MAY have one or more EDNS options RFC6891 in the request." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] A "dns api clients using the dns wire format may have one or more edns options rfc6891 in the request." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "The media type is ""application/dns-message""." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] A "the media type is ""application/dns-message""." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "The data payload for the application/dns-message media type is a single message of the DNS on-the-wire format defined in section 4.2.1 of RFC1035. The format was originally for DNS over UDP. Although RFC1035 says ""Messages carried by UDP are restricted to 512 bytes"", that was later updated by RFC6891. This media type restricts the maximum size of the DNS message to 65535 bytes. Note that the wire format used in this media type is different than the wire format used in RFC7858 (which uses the format defined in section 4.2.2 of RFC1035)." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest 7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] B "the data payload for the application/dns-message media type is a single message of the dns on-the-wire format defined in section 4.2.1 of rfc1035. the format was originally for dns over udp. although rfc1035 says ""messages carried by udp are restricted to 512 bytes"", that was later updated by rfc6891. this media type restricts the maximum size of the dns message to 65535 bytes. note that the wire format used in this media type is different than the wire format used in rfc7858 (which uses the format defined in section 4.2.2 of rfc1035)." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "DNS API clients using this media type MAY have one or more EDNS options RFC6891 in the request. DNS API servers using this media type MUST ignore the value given for the EDNS UDP payload size in DNS requests." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest 7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] B "dns api clients using this media type may have one or more edns options rfc6891 in the request. dns api servers using this media type must ignore the value given for the edns udp payload size in dns requests." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "When using the GET method, the data payload for this media type MUST be encoded with base64url RFC4648 and then provided as a variable named ""dns"" to the URI Template expansion. Padding characters for base64url MUST NOT be included." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest 363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] B "when using the get method, the data payload for this media type must be encoded with base64url rfc4648 and then provided as a variable named ""dns"" to the uri template expansion. padding characters for base64url must not be included." 196 {71, 72, 76, 77, 48, 55} {76: 0, 48: 1, 72: 1, 77: 1, 55: 1, 71: 2} {196: 0, 131: 1, 144: 1, 122: 1, 199: 1, 142: 2} "When using the POST method, the data payload for this media type MUST NOT be encoded and is used directly as the HTTP message body." " has given this thumbs up out of band This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" special meta QTYPEs "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" EDNS UDP Payload size cache issue "" This issue is here just so we do n't lose it ; the discussion is happening on the mailing list under the subject line that contains "" TC and message size "" ." 1ab35d145b8e51752eeedea22609877d4e6686e9 1ab35d145b8e51752eeedea22609877d4e6686e9 draft-ietf-doh-dns-over-https-latest 0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', '0cb5dc751b9985df24a7627addb62ff23eacb8378f51718c4f388599c7ff91d8', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347', 'f7ded83ddf7e1daaca7527f8e5f55529db4ed1a2b144189f1debd410801febe0', 'c0d6b8560a2aded13eecc60da9cb7827ebffd3be68f959cdabbb3ec7d92f9052', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e', '7a87536ad40329969d29b5f87d3aa700bc46441683bb569e95f1fda8b70d266c', '363b2ba0f31601867c86fe0a04332dec680271cdd158c797ee555e1bb3444c68', '7e2f630838c95b6708f63f8972aedcc9ef1a06d528a27f7cca48d1b97aa31463'] B "when using the post method, the data payload for this media type must not be encoded and is used directly as the http message body." 188 {88, 25, 89} {88: 0, 89: 1, 25: 1} {188: 0, 174: 1, 186: 1} "Before using a DNS API server for DNS resolution, the client MUST establish that the HTTP request URI is a trusted service for the DOH query, in other words, a DNS API client MUST only use a DNS API server that is configured as trustworthy." "This is similar to but , in my opinion , slightly different . The Introduction text says : I guess this implies integrity and confidentiality over the wire . For example DoH does n't guarantee confidentiality from the DoH server learning the content of the DNS query . Ok so it 's pretty clear that DoH focuses on the wire . Someone could argue it 's obvious but the document actually talks about it . Then in Selection of DNS API Server the text says : And because it 's a MUST , not trusting the server violates the DoH specification and hence its benefits ? made the comment that trust is a client issue ( ) and I agree but there are different kinds of trust against different threats . I think we should tell clients what kind of trust they need to establish . Clients might trust the TLS connection to the server because , e.g. , it uses a strong CipherSuite and they can verify the server 's identity . At the same time clients might not trust the server to not profile them based on their queries and sell that information to advertisers . ( In this case clients might obscure their source IP address ) Clients might also not trust that the server is being diligent in checking the RRSIG records for every A record it gets from upstream servers . ( In this case clients will do the DNSSEC checks themselves ) In other words , clients may have a trusted connection to the server but at the same time might treat the server 's application logic as untrusted . What should the client do according to the document ? I think the answer is they should use the server because DoH protects queries on the wire and from that aspect the server is trustworthy . But that 's not clear by the current language which seems to imply absolute trust for everything . I would propose that the text is updated to say : I am uncomfortable with this language because it implies some things about how an HTTPS resource is fetched that are defined elsewhere . Heck , with a good cache there might not even be a connection . Let 's not go there . The existing text : "" Before using a DNS API server for DNS resolution , the client MUST establish that the HTTP request URI is a trusted service for the DOH query , in other words , a DNS API client MUST only use a DNS API server that is configured as trustworthy . "" It really tries to say the same thing twice ( "" in other words "" ) - which may cause confusion . They can be blended to be : "" A DNS API client selects the URI , and by extension DNS API server , used for resolution via configuration . "" This does remove an appropriate ( imo ) MUST , but the next sentence contains a MUST NOT that covers the important part . ? This text is much better - I support PR Only minor nit is that I might ask it says ' direct configuration ' to distinguish it from any type of dynamic configuration ?" 3e3dc83c5615c240931fac48f5a2a0d568673af9 3e3dc83c5615c240931fac48f5a2a0d568673af9 draft-ietf-doh-dns-over-https-latest c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff ['c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff', '7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d', '3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/issues/117#issuecomment-375622717) A "before using a dns api server for dns resolution, the client must establish that the http request uri is a trusted service for the doh query, in other words, a dns api client must only use a dns api server that is configured as trustworthy." 188 {88, 25, 89} {88: 0, 89: 1, 25: 1} {188: 0, 174: 1, 186: 1} "A client MUST NOT use a DNS API server simply because it was discovered, or because the client was told to use the DNS API server by an untrusted party." "This is similar to but , in my opinion , slightly different . The Introduction text says : I guess this implies integrity and confidentiality over the wire . For example DoH does n't guarantee confidentiality from the DoH server learning the content of the DNS query . Ok so it 's pretty clear that DoH focuses on the wire . Someone could argue it 's obvious but the document actually talks about it . Then in Selection of DNS API Server the text says : And because it 's a MUST , not trusting the server violates the DoH specification and hence its benefits ? made the comment that trust is a client issue ( ) and I agree but there are different kinds of trust against different threats . I think we should tell clients what kind of trust they need to establish . Clients might trust the TLS connection to the server because , e.g. , it uses a strong CipherSuite and they can verify the server 's identity . At the same time clients might not trust the server to not profile them based on their queries and sell that information to advertisers . ( In this case clients might obscure their source IP address ) Clients might also not trust that the server is being diligent in checking the RRSIG records for every A record it gets from upstream servers . ( In this case clients will do the DNSSEC checks themselves ) In other words , clients may have a trusted connection to the server but at the same time might treat the server 's application logic as untrusted . What should the client do according to the document ? I think the answer is they should use the server because DoH protects queries on the wire and from that aspect the server is trustworthy . But that 's not clear by the current language which seems to imply absolute trust for everything . I would propose that the text is updated to say : I am uncomfortable with this language because it implies some things about how an HTTPS resource is fetched that are defined elsewhere . Heck , with a good cache there might not even be a connection . Let 's not go there . The existing text : "" Before using a DNS API server for DNS resolution , the client MUST establish that the HTTP request URI is a trusted service for the DOH query , in other words , a DNS API client MUST only use a DNS API server that is configured as trustworthy . "" It really tries to say the same thing twice ( "" in other words "" ) - which may cause confusion . They can be blended to be : "" A DNS API client selects the URI , and by extension DNS API server , used for resolution via configuration . "" This does remove an appropriate ( imo ) MUST , but the next sentence contains a MUST NOT that covers the important part . ? This text is much better - I support PR Only minor nit is that I might ask it says ' direct configuration ' to distinguish it from any type of dynamic configuration ?" 3e3dc83c5615c240931fac48f5a2a0d568673af9 3e3dc83c5615c240931fac48f5a2a0d568673af9 draft-ietf-doh-dns-over-https-latest 7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d ['c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff', '7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d', '3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/issues/117#issuecomment-375622717) A "a client must not use a dns api server simply because it was discovered, or because the client was told to use the dns api server by an untrusted party." 188 {88, 25, 89} {88: 0, 89: 1, 25: 1} {188: 0, 174: 1, 186: 1} "A DNS API client uses configuration to select the URI, and thus the DNS API server, used for resolution. A client MUST NOT use a DNS API server simply because it was discovered, or because the client was told to use the DNS API server by an untrusted party." "This is similar to but , in my opinion , slightly different . The Introduction text says : I guess this implies integrity and confidentiality over the wire . For example DoH does n't guarantee confidentiality from the DoH server learning the content of the DNS query . Ok so it 's pretty clear that DoH focuses on the wire . Someone could argue it 's obvious but the document actually talks about it . Then in Selection of DNS API Server the text says : And because it 's a MUST , not trusting the server violates the DoH specification and hence its benefits ? made the comment that trust is a client issue ( ) and I agree but there are different kinds of trust against different threats . I think we should tell clients what kind of trust they need to establish . Clients might trust the TLS connection to the server because , e.g. , it uses a strong CipherSuite and they can verify the server 's identity . At the same time clients might not trust the server to not profile them based on their queries and sell that information to advertisers . ( In this case clients might obscure their source IP address ) Clients might also not trust that the server is being diligent in checking the RRSIG records for every A record it gets from upstream servers . ( In this case clients will do the DNSSEC checks themselves ) In other words , clients may have a trusted connection to the server but at the same time might treat the server 's application logic as untrusted . What should the client do according to the document ? I think the answer is they should use the server because DoH protects queries on the wire and from that aspect the server is trustworthy . But that 's not clear by the current language which seems to imply absolute trust for everything . I would propose that the text is updated to say : I am uncomfortable with this language because it implies some things about how an HTTPS resource is fetched that are defined elsewhere . Heck , with a good cache there might not even be a connection . Let 's not go there . The existing text : "" Before using a DNS API server for DNS resolution , the client MUST establish that the HTTP request URI is a trusted service for the DOH query , in other words , a DNS API client MUST only use a DNS API server that is configured as trustworthy . "" It really tries to say the same thing twice ( "" in other words "" ) - which may cause confusion . They can be blended to be : "" A DNS API client selects the URI , and by extension DNS API server , used for resolution via configuration . "" This does remove an appropriate ( imo ) MUST , but the next sentence contains a MUST NOT that covers the important part . ? This text is much better - I support PR Only minor nit is that I might ask it says ' direct configuration ' to distinguish it from any type of dynamic configuration ?" 3e3dc83c5615c240931fac48f5a2a0d568673af9 3e3dc83c5615c240931fac48f5a2a0d568673af9 draft-ietf-doh-dns-over-https-latest 3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41 ['c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff', '7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d', '3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/issues/117#issuecomment-375622717) B "a dns api client uses configuration to select the uri, and thus the dns api server, used for resolution. a client must not use a dns api server simply because it was discovered, or because the client was told to use the dns api server by an untrusted party." 187 {49, 84, 85} {84: 0, 85: 1, 49: 2} {187: 0, 116: 1, 89: 2} "A client can use DNS over HTTPS as one of multiple mechanisms to obtain DNS data. If a client of this protocol encounters an HTTP error after sending a DNS query, and then falls back to a different DNS retrieval mechanism, doing so can weaken the privacy and authenticity expected by the user of the client." "this paragraph is true .. but its a tad confusing and I do n't think really describes anything germane to doh . WFM to just remove this text" 1fd890ce6bc4ad2c35af511ddb32db22a663263d 1fd890ce6bc4ad2c35af511ddb32db22a663263d draft-ietf-doh-dns-over-https-latest be312f4e116d747b2945a7685e0a3846619c91137fc5169e4f16adc33e4b568e ['be312f4e116d747b2945a7685e0a3846619c91137fc5169e4f16adc33e4b568e'] A "a client can use dns over https as one of multiple mechanisms to obtain dns data. if a client of this protocol encounters an http error after sending a dns query, and then falls back to a different dns retrieval mechanism, doing so can weaken the privacy and authenticity expected by the user of the client." 186 {88, 25, 89} {25: 0, 88: 1, 89: 2} {186: 0, 188: 1, 174: 2} "A DNS API client uses configuration to select the URI, and thus the DNS API server, used for resolution. A client MUST NOT use a DNS API server simply because it was discovered, or because the client was told to use the DNS API server by an untrusted party." "clarify that server identity authentication is part of . Thanks - this addresses my request ." 57271adf57bb88ebd7b9cfff89541de0214241b4 57271adf57bb88ebd7b9cfff89541de0214241b4 draft-ietf-doh-dns-over-https-latest 3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41 ['3669db81c7d8ddbdea2e2b5f40feeb573123bf93288aee968c12e8ea027a903f', '3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41'] A "a dns api client uses configuration to select the uri, and thus the dns api server, used for resolution. a client must not use a dns api server simply because it was discovered, or because the client was told to use the dns api server by an untrusted party." 186 {88, 25, 89} {25: 0, 88: 1, 89: 2} {186: 0, 188: 1, 174: 2} "A DNS API client uses configuration to select the URI, and thus the DNS API server, used for resolution. A client MUST NOT use a DNS API server simply because it was discovered, or because the client was told to use the DNS API server by an untrusted party. RFC2818 defines how HTTPS verifies the server's identity." "clarify that server identity authentication is part of . Thanks - this addresses my request ." 57271adf57bb88ebd7b9cfff89541de0214241b4 57271adf57bb88ebd7b9cfff89541de0214241b4 draft-ietf-doh-dns-over-https-latest 3669db81c7d8ddbdea2e2b5f40feeb573123bf93288aee968c12e8ea027a903f ['3669db81c7d8ddbdea2e2b5f40feeb573123bf93288aee968c12e8ea027a903f', '3e0a70ffe1884722116154f0ba6b9d8043e5460108a116f615e0b10f65fdba41'] B "a dns api client uses configuration to select the uri, and thus the dns api server, used for resolution. a client must not use a dns api server simply because it was discovered, or because the client was told to use the dns api server by an untrusted party. rfc2818 defines how https verifies the server's identity." 178 {100, 114, 24, 27, 28} {100: 0, 114: 1, 28: 2, 27: 3, 24: 4} {178: 0, 176: 1, 174: 2, 172: 3, 157: 4} "In the absence of DNSSEC information about the authenticity of responses a DNS API server can give a client invalid data in responses. A client MUST NOT use arbitrary DNS API servers. Instead, a client MUST only use DNS API servers specified using mechanisms such as explicit configuration. This does not guarantee protection against invalid data but reduces the risk." "Remove double use of "" responses "" . ok by me . assigning to " 7f8926486003bcfbb077a52a75aef9c663117a58 7f8926486003bcfbb077a52a75aef9c663117a58 draft-ietf-doh-dns-over-https-latest cf5b962224699dab07415aff6359e550886ff16c360e4b60b25c3e70b4fe4ac0 ['cf5b962224699dab07415aff6359e550886ff16c360e4b60b25c3e70b4fe4ac0', '88b5c6e2f1fff5f01f4219cb0f673dcfb3dd031ce73e95d9fdedc080245caa92'] A "in the absence of dnssec information about the authenticity of responses a dns api server can give a client invalid data in responses. a client must not use arbitrary dns api servers. instead, a client must only use dns api servers specified using mechanisms such as explicit configuration. this does not guarantee protection against invalid data but reduces the risk." 178 {100, 114, 24, 27, 28} {100: 0, 114: 1, 28: 2, 27: 3, 24: 4} {178: 0, 176: 1, 174: 2, 172: 3, 157: 4} "In the absence of DNSSEC information, a DNS API server can give a client invalid data in response to a DNS query. A client MUST NOT use arbitrary DNS API servers. Instead, a client MUST only use DNS API servers specified using mechanisms such as explicit configuration. This does not guarantee protection against invalid data but reduces the risk." "Remove double use of "" responses "" . ok by me . assigning to " 7f8926486003bcfbb077a52a75aef9c663117a58 7f8926486003bcfbb077a52a75aef9c663117a58 draft-ietf-doh-dns-over-https-latest 88b5c6e2f1fff5f01f4219cb0f673dcfb3dd031ce73e95d9fdedc080245caa92 ['cf5b962224699dab07415aff6359e550886ff16c360e4b60b25c3e70b4fe4ac0', '88b5c6e2f1fff5f01f4219cb0f673dcfb3dd031ce73e95d9fdedc080245caa92'] B "in the absence of dnssec information, a dns api server can give a client invalid data in response to a dns query. a client must not use arbitrary dns api servers. instead, a client must only use dns api servers specified using mechanisms such as explicit configuration. this does not guarantee protection against invalid data but reduces the risk." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "A DNS API client may utilize a hierarchy of caches that include both HTTP and DNS specific caches. HTTP cache entries may be bypassed with HTTP mechanisms such as the ""no-cache"" Cache-Control directive; however DNS caches do not have a similar mechanism." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "a dns api client may utilize a hierarchy of caches that include both http and dns specific caches. http cache entries may be bypassed with http mechanisms such as the ""no-cache"" cache-control directive; however dns caches do not have a similar mechanism." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "The Answer section of a DNS response can contain zero or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset has Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs, although it is only possible for the HTTP response to have a single freshness lifetime. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the RRset with the smallest TTL in the Answer section of the response. Specifically, the HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS resource records in the Answer section reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "the answer section of a dns response can contain zero or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset has time to live (ttl) freshness information. different rrsets in the answer section can have different ttls, although it is only possible for the http response to have a single freshness lifetime. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the rrset with the smallest ttl in the answer section of the response. specifically, the http freshness lifetime should be set to expire at the same time any of the dns resource records in the answer section reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns resource record with the smallest ttl in the response." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "If the DNS response has no records in the Answer section, and the DNS response has an SOA record in the Authority section, the response freshness lifetime MUST NOT be greater than the MINIMUM field from that SOA record. (See RFC2308.) Otherwise, the HTTP response MUST set a freshness lifetime (RFC7234 Section 4.2) of 0 by using a mechanism such as the ""no-cache"" Cache-Control directive (RFC7234 Section 5.2.1.4)." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "if the dns response has no records in the answer section, and the dns response has an soa record in the authority section, the response freshness lifetime must not be greater than the minimum field from that soa record. (see rfc2308.) otherwise, the http response must set a freshness lifetime (rfc7234 section 4.2) of 0 by using a mechanism such as the ""no-cache"" cache-control directive (rfc7234 section 5.2.1.4)." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "A DNS API client that receives a response without an explicit freshness lifetime MUST NOT assign that response a heuristic freshness (RFC7234 Section 4.2.2.) greater than that indicated by the DNS Record with the smallest TTL in the response." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "a dns api client that receives a response without an explicit freshness lifetime must not assign that response a heuristic freshness (rfc7234 section 4.2.2.) greater than that indicated by the dns record with the smallest ttl in the response." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "A DOH response that was previously stored in an HTTP cache will contain the RFC7234 Age response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response. DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "a doh response that was previously stored in an http cache will contain the rfc7234 age response header indicating the elapsed time between when the entry was placed in the http cache and the current doh response. dns api clients should subtract this time from the dns ttl in each rrset to determine the time remaining on the ttl." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "HTTP revalidation (e.g., via If-None-Match request headers) of cached DNS information may be of limited value to DOH as revalidation provides only a bandwidth benefit and DNS transactions are normally latency bound. Furthermore, the HTTP response headers that enable revalidation (such as ""Last-Modified"" and ""Etag"") are often fairly large when compared to the overall DNS response size, and have a variable nature that creates constant pressure on the HTTP/2 compression dictionary RFC7541. Other types of DNS data, such as zone transfers, may be larger and benefit more from revalidation. DNS API servers may wish to consider whether providing these validation enabling response headers is worthwhile." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "http revalidation (e.g., via if-none-match request headers) of cached dns information may be of limited value to doh as revalidation provides only a bandwidth benefit and dns transactions are normally latency bound. furthermore, the http response headers that enable revalidation (such as ""last-modified"" and ""etag"") are often fairly large when compared to the overall dns response size, and have a variable nature that creates constant pressure on the http/2 compression dictionary rfc7541. other types of dns data, such as zone transfers, may be larger and benefit more from revalidation. dns api servers may wish to consider whether providing these validation enabling response headers is worthwhile." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "The ""stale-while-revalidate"" and ""stale-if-error"" Cache-Control directives (RFC5861) could be well suited to a DOH implementation when allowed by server policy. Those mechanisms allow a client, at the server's discretion, to reuse a cache entry that is no longer fresh." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "the ""stale-while-revalidate"" and ""stale-if-error"" cache-control directives (rfc5861) could be well suited to a doh implementation when allowed by server policy. those mechanisms allow a client, at the server's discretion, to reuse a cache entry that is no longer fresh." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "All HTTP servers, including DNS API servers, need to consider cache interaction when they generate responses that are not globally valid. For instance, if a DNS API server customized a response based on the client's identity then it would not want to globally allow reuse of that response. This could be accomplished through a variety of HTTP techniques such as a Cache-Control max-age of 0, or perhaps by the Vary response header." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" A "all http servers, including dns api servers, need to consider cache interaction when they generate responses that are not globally valid. for instance, if a dns api server customized a response based on the client's identity then it would not want to globally allow reuse of that response. this could be accomplished through a variety of http techniques such as a cache-control max-age of 0, or perhaps by the vary response header." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "A DOH exchange can pass through a hierarchy of caches that include both HTTP and DNS specific caches. These caches may exist beteen the DNS API server and client, or on the DNS API client itself. HTTP caches are by design generic; that is, they do not understand this protocol. Even if a DNS API client has modified its cache implementation to be aware of DOH semantics, it does not follow that all upstream caches (for example, inline proxies, server-side gateways and Content Delivery Networks) will be." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "a doh exchange can pass through a hierarchy of caches that include both http and dns specific caches. these caches may exist beteen the dns api server and client, or on the dns api client itself. http caches are by design generic; that is, they do not understand this protocol. even if a dns api client has modified its cache implementation to be aware of doh semantics, it does not follow that all upstream caches (for example, inline proxies, server-side gateways and content delivery networks) will be." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "As a result, DNS API servers need to carefully consider the HTTP caching metadata they send in response to GET requests (POST requests are not cacheable unless specific response headers are sent; this is not widely implemented, and not advised for DOH)." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "as a result, dns api servers need to carefully consider the http caching metadata they send in response to get requests (post requests are not cacheable unless specific response headers are sent; this is not widely implemented, and not advised for doh)." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "In particular, DNS API servers SHOULD assign an explicit freshness lifetime (RFC7234 Section 4.2) so that the DNS API client is more likely to use fresh DNS data. This requirement is due to HTTP caches being able to assign their own heuristic freshness (such as that described in RFC7234 Section 4.2.2), which would take control of the cache contents out of the hands of the DNS API server." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "in particular, dns api servers should assign an explicit freshness lifetime (rfc7234 section 4.2) so that the dns api client is more likely to use fresh dns data. this requirement is due to http caches being able to assign their own heuristic freshness (such as that described in rfc7234 section 4.2.2), which would take control of the cache contents out of the hands of the dns api server." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "The assigned freshness lifetime of a DOH HTTP response SHOULD be the smallest TTL in the Answer section of the DNS response. For example, if a HTTP response carries three RRsets with TTLs of 30, 600, and 300, the HTTP freshness lifetime should be 30 seconds (which could be specified as ""Cache-Control: max-age=30""). The assigned freshness lifetime MUST NOT be greater than the smallest TTL in the Answer section of the DNS response. This requirement helps assure that none of the RRsets contained in a DNS response are served stale from an HTTP cache." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "the assigned freshness lifetime of a doh http response should be the smallest ttl in the answer section of the dns response. for example, if a http response carries three rrsets with ttls of 30, 600, and 300, the http freshness lifetime should be 30 seconds (which could be specified as ""cache-control: max-age=30""). the assigned freshness lifetime must not be greater than the smallest ttl in the answer section of the dns response. this requirement helps assure that none of the rrsets contained in a dns response are served stale from an http cache." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "If the DNS response has no records in the Answer section, and the DNS response has an SOA record in the Authority section, the response freshness lifetime MUST NOT be greater than the MINIMUM field from that SOA record (see RFC2308)." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "if the dns response has no records in the answer section, and the dns response has an soa record in the authority section, the response freshness lifetime must not be greater than the minimum field from that soa record (see rfc2308)." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "The stale-while-revalidate and stale-if-error Cache-Control directives (RFC5861) could be well suited to a DOH implementation when allowed by server policy. Those mechanisms allow a client, at the server's discretion, to reuse a cache entry that is no longer fresh. In such a case, the client reuses all of a cached entry, or none of it." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "the stale-while-revalidate and stale-if-error cache-control directives (rfc5861) could be well suited to a doh implementation when allowed by server policy. those mechanisms allow a client, at the server's discretion, to reuse a cache entry that is no longer fresh. in such a case, the client reuses all of a cached entry, or none of it." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "DNS API servers also need to consider caching when generating responses that are not globally valid. For instance, if a DNS API server customizes a response based on the client's identity, it would not want to allow global reuse of that response. This could be accomplished through a variety of HTTP techniques such as a Cache- Control max-age of 0, or by using the Vary response header (RFC7231 Section 7.1.4) to establish a secondary cache key (RFC7234 Section 4.1)." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "dns api servers also need to consider caching when generating responses that are not globally valid. for instance, if a dns api server customizes a response based on the client's identity, it would not want to allow global reuse of that response. this could be accomplished through a variety of http techniques such as a cache- control max-age of 0, or by using the vary response header (rfc7231 section 7.1.4) to establish a secondary cache key (rfc7234 section 4.1)." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "DNS API clients MUST account for the Age response header's value (RFC7234) when calculating the DNS TTL of a response. For example, if a RRset is received with a DNS TTL of 600, but the Age header indicates that the response has been cached for 250 seconds, the remaining lifetime of the RRset is 350 seconds." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "dns api clients must account for the age response header's value (rfc7234) when calculating the dns ttl of a response. for example, if a rrset is received with a dns ttl of 600, but the age header indicates that the response has been cached for 250 seconds, the remaining lifetime of the rrset is 350 seconds." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "DNS API clients can request an uncached copy of a response by using the ""no-cache"" request cache control directive (RFC7234, Section 5.2.1.4) and similar controls. Note that some caches might not honor these directives, either due to configuration or interaction with traditional DNS caches that do not have such a mechanism." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "dns api clients can request an uncached copy of a response by using the ""no-cache"" request cache control directive (rfc7234, section 5.2.1.4) and similar controls. note that some caches might not honor these directives, either due to configuration or interaction with traditional dns caches that do not have such a mechanism." 177 {0, 1, 74, 43, 62} {1: 0, 0: 1, 74: 1, 43: 1, 62: 1} {177: 0, 199: 1, 56: 1, 154: 1} "HTTP conditional requests (RFC7232) may be of limited value to DOH, as revalidation provides only a bandwidth benefit and DNS transactions are normally latency bound. Furthermore, the HTTP response headers that enable revalidation (such as ""Last-Modified"" and ""Etag"") are often fairly large when compared to the overall DNS response size, and have a variable nature that creates constant pressure on the HTTP/2 compression dictionary RFC7541. Other types of DNS data, such as zone transfers, may be larger and benefit more from revalidation." "In reading this a second time through , I wonder what value there is in matching the freshness lifetime to the TTL ? The implicit assumption here is that it is important that a cache never serve a response that includes an expired RRset . This simplifies the implementation of a client , but it also means that it tends to make client 's stupid . That is , they might not use RRsets that are still valid from the perspective of their TTL , simply because they are no longer in cache . It might be better to have the DNS code that uses the RRset being responsible for checking the TTL . HTTP will ensure that the time that the RRset was generated is faithfully recorded and transferred with the response , so that should be enough . It potentially allows use of old responses for RRsets that are still valid . Finally , the mention of stale - while - revalidate and stale - if - error actually suggests that this requirement is more . These operate outside of the freshness lifetime . I think it 's important that DOH be usable with standard ( non - DOH - aware ) HTTP caches . Allowing the cache lifetime to exceed the shortest TTL might make it impossible to use DOH through such a cache : the client can not flush the cache , and the cache entry might not expire until after its longest TTL . +1 to response . Clients can always set Cache - Control : no - cache and avoid the cache . The point here is that the TTL is the authoritative value . Insisting on HTTP matching that both loses fidelity ( multiple RRsets have different TTLs ) and encourages lazy implementation of clients . Agree with everyone . I would say that a server / parser could in that case imagine to set a "" retry - after:60 "" HTTP header ( corresponding to the lower TTL in this example RR ) ; when TTL < RR situation verifies ( simple calculation , t0 plus lowest values in RR ) , a "" vary : retry - after "" header could control the re - freshing of object ( aka , force refresh sometimes but do still do cache in the meantime ) . By the way , some ISP 's DNS used to generally override all TTL that were lower than 60 seconds , used to be even 300 seconds . halfway between some cache and good reachability / redundancy would be good enough ( TTL real value is to support / improve the service continuity / validity . TTL mission is not guessing how long a service would keep being available , or running ) . Some services as Skype would get such cache crazy , due to 5 to 30 seconds .... a choice must be done somewhere between precision and performances .. An average HTTP cache lasts between minutes and days , say everywhere from 1 to 8 hours in proxy administration praxis , 24 hours . 24 hours in DNS terms is unacceptable . How accurately will the HTML programmer choose such blind - validity duration ? In some ways , choosing HTTP validity exactly the same as the choiche of TTL for a given record , a pure bet ! ! Caching is one of the pillars of DOH in terms of speed and HTTP - friendliness . As one of the goals is ( i believe ) creating and populating a global valid cache through HTTP ( as alternative or complement of existing DNS cache ) , doing no - cache is also a bit "" against "" the principle . but still a feature for privacy amateurs . I see value in HTTP cache also in case of DDoS ; HTTP / DOH cache could represent a good fallback / backup of DNS data anyway , and HTTP is proved more resilent to flooding . One of the goals of DOH is to leverage use of existing infrastructures ! et In that case , setting is the only way a client can make sure that it actually gets non - expired answers . This creates a strong incentive , in both performance and complexity , to set on all DOH requests , defeating HTTP caching altogether . Yikes ! I favor keeping the current arrangement . A server that wishes to avoid the crux of the issue ( i.e. that HTTP has one TTL , but the DNS response is a set of TTLs ) can decompose response into multiple HTTP responses using HTTP/2 push each with their own TTLs . That seems like the right approach . I have a pretty strong inclination that the principle of least surprise says data in a successful HTTP response can always be used one time ( but not reused ) even if you do n't implement any caching logic . I know your suggestion does n't technically infringe on that - its more of a requirement on the processing of the response- but it seems like a much more fragile alignment . and I think is on to something important with cache interop . My .02 - I think that the advice in the current draft is good -- but it 's just advice , and should n't have RFC2119 requirements associated with it . From the current markdown : Under what conditions can this SHOULD be violated ? What is a client to do when it encounters a violation ? Suggest "" Specifically , the HTTP freshness lifetime should generally be set ... "" As above , there may be cases where this is desirable . Suggest "" Be aware that if the response freshness lifetime is greater than that indicated by the DNS resource record with the smallest TTL in the response , HTTP caches can store those records for longer than they are valid for ; clients can request a fresh HTTP response ( e.g. , using the Cache - Control : no - cache request directive ) in this situation , but this can add latency . "" Question : should we say anything about whether a DOH client can use a stale HTTP response , and the conditions under which it should ? As above . This is really vague for a MUST . implies ; is that required ? Also , what is the otherwise applying to -- is this all responses that have no records in the answer section and no records in the DNS response ? Many DNS API clients wo n't have any control over the heuristics that their ( or an upstream ) cache applies ; this is n't an enforceable requirement . beat me to it . I think that we 're making suggestions , not interoperability requirements . I would prefer to have no freshness information ( and rely on heuristics ) than to recommend marking things uncacheable . I think that the rules are similar to using ( and inflating ) TTL . Saying nothing might be easiest . Under no circumstances does an HTTP client have to expect that a server may serve it an expired response out of cache , and recover by retrying with caching disabled . Under no circumstance does a DNS client have to expect that a recursive may serve it an expired response , and recover by bypassing the recursive cache somehow . There 's a good reason for this : this architecture is complex , adds latency , and encourages the client to abandon caching rather than accept the complexity and retry latency . I do not think we can reasonably accept a specification that puts DOH clients in this position . To be clear , this is an interoperability question . A server that adopts a longer - than - min - TTL policy and a simple client ( e.g. any existing client ) are not interoperable . longer - than - min - TTL is not significantly different to a DNS server extending the TTL . It 's not advisable , but it still happens . The difference is that HTTP makes this visible , where DNS does not . A client that makes a request can reasonably use the response without checking either freshness lifetime or TTL . The current doc states "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" I am guessing that this and any other similar text should refer to RRSet not RR . I really hate the "" A DOH response that was previously stored in an HTTP cache will contain the response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response . DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL . "" in DNS that is the job of the cache not the client . Or do you want to support DNS unaware HTTP caches ? yes - that 's part of using http as the substrate .. I agree : in both cases , it is a protocol violation . The fact that some recursive resolvers fail to implement the DNS protocol does not mean that we need to endorse their incorrect behavior . If they want to violate RFC 1035 , they can equally well violate DOH . The fact that these violations usually mostly work in practice is all the more reason to be clear about the designed behavior , because people who get it wrong wo n't notice for a long time . Perhaps we should state this normatively . Hear hear . Exactly the point I was about to make . Please take this all to the mailing list . I have ( again ) opened up a topic there . I 'm just reopening this because I use the github issues list as an editorial punchlist - not a comment about where to discuss :) OK , but the rest of you : please do n't comment here , but please do comment on the list . Thanks ! that is answered in the next sentence : "" The response freshness lifetime MUST NOT be greater than that indicated by the DNS resource record with the smallest TTL in the response . "" its basically saying you SHOULD make HTTP and DNS match , but if you ca n't do that ( because perhaps your code module ca n't fully parse DNS ) you can do something like - age = 0 I will soon commit as my best interpretation of the discussion ( here and on the mailling list ) It accepts most of 's changes but preserves the normative language and adds back a couple of small things that were the result of previous explicit working group requests ( e.g. noting that cache busting is not something that can be done easily to a DNS - only cache ) . Seems mostly good , but see for my edits ." 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 9f33a4b7b7f5d0888bfa5cf010abb5067e07ea31 draft-ietf-doh-dns-over-https-latest 11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3 ['00e8b9067775e21458c180312eaaa9da6322e2bb8df4ca6c41427a3d70549a91', '9131568c1fd09fe86458e83583c25dcbcd9f4d8c04d2d3be4ba671c585457ded', '4578e42c8f00e4d318049b7345822013282d1c78a7e37adc65fe18afa80449c8', 'ed7705d333a695a428495d0507fd620211eb5ec8ba387fbc49b192bb6ad1747e', '2d16a247408a9f6473e9e983b5aa82f029def5c276e99503a1fe78a08c77f5b9', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074', '0f15ebfabb9e2fd537f6d39e6b86995cbac9cbf9af5b395700998d786918341d', '2d9e11c610efc7b27c4a7636429cf7ddcea921d8729e643199eac72a303058d4', 'aaa939e57959431cdf3e66555d3179bac736a72feaba81c3d82218bd57336933', '82bb33fa33da082d2db7e514deacbe0eef4e5336c16fe11819f0bf7b56ab67b2', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', '902535e96edfea9955596057261edda3c0d612064760c0a94a74954f85e428dd', '5f36651911ece9f3c2c94b647c66a22ee391eea832b218c2bafbb08866eec17a', '11412b329bb5c13cc4e86b978d4d61c1b0f0055b1439f2f23045e4c0d3c8a0a3', 'e8996603438573d7326a804c59831dc68a3e41cbbe9a63b5f439dbc87499aea7', '90efcf9e6310d314bf29147add472ab914abf921a7427af4e0a9d67d8a88346b', 'ca1aae5a0600e4afe2fadb6ccc28639fea99d214eb5cbf5350cf7dc98f0383ba', '22fa748ba808700b5cf445b1d564677ebe3205515d6fb736ed3fcd554a1ca87e'] "https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Retry-After https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Vary" B "http conditional requests (rfc7232) may be of limited value to doh, as revalidation provides only a bandwidth benefit and dns transactions are normally latency bound. furthermore, the http response headers that enable revalidation (such as ""last-modified"" and ""etag"") are often fairly large when compared to the overall dns response size, and have a variable nature that creates constant pressure on the http/2 compression dictionary rfc7541. other types of dns data, such as zone transfers, may be larger and benefit more from revalidation." 176 {119, 87} {119: 0, 87: 1} {176: 0, 199: 1} "The HTTPS connection provides transport security for the interaction between the DNS API server and client, but does not provide the response integrity of DNS data provided by DNSSEC. DNSSEC and DOH are independent and fully compatible protocols, each solving different problems. The use of one does not diminish the need nor the usefulness of the other. It is the choice of a client to either perform full DNSSEC validation of answers or to treat answers from a DNS API server as DNSSEC-authenticated, and this choice may be affected by the response media type." "… raft - ietf - doh - dns - over - https / commit/#commitcomment- and that this wording is much clearer . Sara , making this pr was on my todo list - thanks for the time saver . sent me an IM about it , so I 'll ga and merge ." 545776370d792c76eb87b656743e6f4a5ce2c26a 545776370d792c76eb87b656743e6f4a5ce2c26a draft-ietf-doh-dns-over-https-latest fa751a2b7be5b78f893f3835e96a77de70cdffbb2900d25911571226d3a0a34c ['fa751a2b7be5b78f893f3835e96a77de70cdffbb2900d25911571226d3a0a34c', '0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/commit/3b2bddf34291dec24094f5cd0c43df024011a20d#commitcomment-28865660 A "the https connection provides transport security for the interaction between the dns api server and client, but does not provide the response integrity of dns data provided by dnssec. dnssec and doh are independent and fully compatible protocols, each solving different problems. the use of one does not diminish the need nor the usefulness of the other. it is the choice of a client to either perform full dnssec validation of answers or to treat answers from a dns api server as dnssec-authenticated, and this choice may be affected by the response media type." 176 {119, 87} {119: 0, 87: 1} {176: 0, 199: 1} "The HTTPS connection provides transport security for the interaction between the DNS API server and client, but does not provide the response integrity of DNS data provided by DNSSEC. DNSSEC and DOH are independent and fully compatible protocols, each solving different problems. The use of one does not diminish the need nor the usefulness of the other. It is the choice of a client to either perform full DNSSEC validation of answers or to trust the DNS API server to do DNSSEC validation and inspect the AD (Authentic Data) bit in the returned message to determine whether an answer was authentic or not. As noted in http-response, different response media types will provide more or less information from a DNS response so this choice may be affected by the response media type." "… raft - ietf - doh - dns - over - https / commit/#commitcomment- and that this wording is much clearer . Sara , making this pr was on my todo list - thanks for the time saver . sent me an IM about it , so I 'll ga and merge ." 545776370d792c76eb87b656743e6f4a5ce2c26a 545776370d792c76eb87b656743e6f4a5ce2c26a draft-ietf-doh-dns-over-https-latest 0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2 ['fa751a2b7be5b78f893f3835e96a77de70cdffbb2900d25911571226d3a0a34c', '0e1df49544514d540cc870c35b8351503063ec5cbf4569fe3911deabbf75f7e2'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/commit/3b2bddf34291dec24094f5cd0c43df024011a20d#commitcomment-28865660 B "the https connection provides transport security for the interaction between the dns api server and client, but does not provide the response integrity of dns data provided by dnssec. dnssec and doh are independent and fully compatible protocols, each solving different problems. the use of one does not diminish the need nor the usefulness of the other. it is the choice of a client to either perform full dnssec validation of answers or to trust the dns api server to do dnssec validation and inspect the ad (authentic data) bit in the returned message to determine whether an answer was authentic or not. as noted in http-response, different response media types will provide more or less information from a dns response so this choice may be affected by the response media type." 176 {100, 114, 24, 27, 28} {114: 0, 100: 1, 28: 1, 27: 2, 24: 3} {176: 0, 178: 1, 174: 1, 172: 2, 157: 3} "In the absence of information about the authenticity of responses, such as DNSSEC, a DNS API server can give a client invalid data in responses. A client MUST NOT use arbitrary DNS API servers. Instead, a client MUST only use DNS API servers specified using mechanisms such as explicit configuration. This does not guarantee protection against invalid data but reduces the risk." "… raft - ietf - doh - dns - over - https / commit/#commitcomment- and that this wording is much clearer . Sara , making this pr was on my todo list - thanks for the time saver . sent me an IM about it , so I 'll ga and merge ." 545776370d792c76eb87b656743e6f4a5ce2c26a 545776370d792c76eb87b656743e6f4a5ce2c26a draft-ietf-doh-dns-over-https-latest f830309268e3712db11e75cc782935a39278ca62dd8c819d555be1200be185a0 ['f830309268e3712db11e75cc782935a39278ca62dd8c819d555be1200be185a0', 'cf5b962224699dab07415aff6359e550886ff16c360e4b60b25c3e70b4fe4ac0'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/commit/3b2bddf34291dec24094f5cd0c43df024011a20d#commitcomment-28865660 A "in the absence of information about the authenticity of responses, such as dnssec, a dns api server can give a client invalid data in responses. a client must not use arbitrary dns api servers. instead, a client must only use dns api servers specified using mechanisms such as explicit configuration. this does not guarantee protection against invalid data but reduces the risk." 176 {100, 114, 24, 27, 28} {114: 0, 100: 1, 28: 1, 27: 2, 24: 3} {176: 0, 178: 1, 174: 1, 172: 2, 157: 3} "In the absence of DNSSEC information about the authenticity of responses a DNS API server can give a client invalid data in responses. A client MUST NOT use arbitrary DNS API servers. Instead, a client MUST only use DNS API servers specified using mechanisms such as explicit configuration. This does not guarantee protection against invalid data but reduces the risk." "… raft - ietf - doh - dns - over - https / commit/#commitcomment- and that this wording is much clearer . Sara , making this pr was on my todo list - thanks for the time saver . sent me an IM about it , so I 'll ga and merge ." 545776370d792c76eb87b656743e6f4a5ce2c26a 545776370d792c76eb87b656743e6f4a5ce2c26a draft-ietf-doh-dns-over-https-latest cf5b962224699dab07415aff6359e550886ff16c360e4b60b25c3e70b4fe4ac0 ['f830309268e3712db11e75cc782935a39278ca62dd8c819d555be1200be185a0', 'cf5b962224699dab07415aff6359e550886ff16c360e4b60b25c3e70b4fe4ac0'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/commit/3b2bddf34291dec24094f5cd0c43df024011a20d#commitcomment-28865660 B "in the absence of dnssec information about the authenticity of responses a dns api server can give a client invalid data in responses. a client must not use arbitrary dns api servers. instead, a client must only use dns api servers specified using mechanisms such as explicit configuration. this does not guarantee protection against invalid data but reduces the risk." 174 {88, 25, 89} {89: 0, 88: 1, 25: 2} {174: 0, 188: 1, 186: 2} "Before using a DNS API server for DNS resolution, the client MUST establish that the HTTP request URI is a trusted service for the DOH query, in other words, a DNS API client MUST only use a DNS API server that is configured as trustworthy." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff ['c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff', '7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d', 'e530b5d00fb7867f455037a4a89bb9ff60885c6c9eeba88d48c850c1b316d1a9'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html B "before using a dns api server for dns resolution, the client must establish that the http request uri is a trusted service for the doh query, in other words, a dns api client must only use a dns api server that is configured as trustworthy." 174 {88, 25, 89} {89: 0, 88: 1, 25: 2} {174: 0, 188: 1, 186: 2} "A client MUST NOT use a DNS API server simply because it was discovered, or because the client was told to use the DNS API server by an untrusted party." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest 7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d ['c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff', '7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d', 'e530b5d00fb7867f455037a4a89bb9ff60885c6c9eeba88d48c850c1b316d1a9'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html B "a client must not use a dns api server simply because it was discovered, or because the client was told to use the dns api server by an untrusted party." 174 {88, 25, 89} {89: 0, 88: 1, 25: 2} {174: 0, 188: 1, 186: 2} "This specification does not extend DNS resolution privileges to URIs that are not recognized by the DNS API client as trusted DNS API servers. As such, use of untrusted servers is out of scope of this document." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest e530b5d00fb7867f455037a4a89bb9ff60885c6c9eeba88d48c850c1b316d1a9 ['c244a38f18efcbc816d12885fc1782f62bd718e168d45728ce85e514397826ff', '7c6ca038d689cd7d944865abd6a4b563663e160134e192933491e3707f1c4c7d', 'e530b5d00fb7867f455037a4a89bb9ff60885c6c9eeba88d48c850c1b316d1a9'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html B "this specification does not extend dns resolution privileges to uris that are not recognized by the dns api client as trusted dns api servers. as such, use of untrusted servers is out of scope of this document." 174 {32, 101, 102, 103, 40, 44, 45, 31} {40: 0, 32: 1, 103: 1, 31: 2, 45: 2, 102: 3, 101: 4, 44: 5} {174: 0, 157: 1, 199: 1, 63: 2, 169: 2, 144: 3, 142: 4, 128: 5} "Before using DOH response data for DNS resolution, the client MUST establish that the HTTP request URI is authorized for the DOH query. For HTTP requests initiated by the DNS API client this authorization is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request. This specification does not extend DNS resolution privileges to URIs that are not recognized by the client as authorized DNS API servers." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest 61088f578d4ba35f0e73790d7f9ad91298a034e71ed8e0ea5514c5041daf65d4 ['61088f578d4ba35f0e73790d7f9ad91298a034e71ed8e0ea5514c5041daf65d4', 'd6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html A "before using doh response data for dns resolution, the client must establish that the http request uri is authorized for the doh query. for http requests initiated by the dns api client this authorization is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request. this specification does not extend dns resolution privileges to uris that are not recognized by the client as authorized dns api servers." 174 {32, 101, 102, 103, 40, 44, 45, 31} {40: 0, 32: 1, 103: 1, 31: 2, 45: 2, 102: 3, 101: 4, 44: 5} {174: 0, 157: 1, 199: 1, 63: 2, 169: 2, 144: 3, 142: 4, 128: 5} "Before using DOH response data for DNS resolution, the client MUST establish that the HTTP request URI may be used for the DOH query. For HTTP requests initiated by the DNS API client this is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest d6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103 ['61088f578d4ba35f0e73790d7f9ad91298a034e71ed8e0ea5514c5041daf65d4', 'd6ba1a6bbea0fe7b7ca98dc2ae80cab111053eb83078fe40ace9acf377e72103'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html B "before using doh response data for dns resolution, the client must establish that the http request uri may be used for the doh query. for http requests initiated by the dns api client this is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request." 174 {100, 114, 24, 27, 28} {28: 0, 114: 1, 27: 1, 100: 2, 24: 2} {174: 0, 176: 1, 172: 1, 178: 2, 157: 2} "In the absence of information about the authenticity of responses, such as DNSSEC, a DNS API server can give a client invalid data in responses. A client MUST NOT authorize arbitrary DNS API servers. Instead, a client MUST specifically authorize DNS API servers using mechanisms such as explicit configuration." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest 3f9d831263782d3076308a307802b6e144ea35b0735a0d2272d113d667cca662 ['3f9d831263782d3076308a307802b6e144ea35b0735a0d2272d113d667cca662', 'f830309268e3712db11e75cc782935a39278ca62dd8c819d555be1200be185a0'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html A "in the absence of information about the authenticity of responses, such as dnssec, a dns api server can give a client invalid data in responses. a client must not authorize arbitrary dns api servers. instead, a client must specifically authorize dns api servers using mechanisms such as explicit configuration." 174 {100, 114, 24, 27, 28} {28: 0, 114: 1, 27: 1, 100: 2, 24: 2} {174: 0, 176: 1, 172: 1, 178: 2, 157: 2} "In the absence of information about the authenticity of responses, such as DNSSEC, a DNS API server can give a client invalid data in responses. A client MUST NOT use arbitrary DNS API servers. Instead, a client MUST only use DNS API servers specified using mechanisms such as explicit configuration. This does not guarantee protection against invalid data but reduces the risk." "Sorry , I may be late to this party ... I 'm not keen on the use of and when discussing which DNS servers a client should choose / trust - those terms are used when discussing technical implementation details , and choosing whether to trust a DNS server is n't that - per RFC2119 : Additionally , without providing an explicit means of determining how and whether a DNS server is trusted , we 're not really helping - we 're just saying that the client MUST make that determination , with no further details . I 'd be fine with replacing them with lower - case non - normative and/or , e.g. : the HTTP request URI is a trusted service for the DOH query ; in other words , a DNS API client should only use a DNS API server that is determined to be trustworthy . because the client was told to use the DNS API server by an untrusted party . not recognized by the DNS API client as trusted DNS API servers . As such , use of untrusted servers is out of scope of this document . its actually not uncommon to leave the ultimate decision about trust out of a protocol definition while still saying that the protocol needs to make such a distinction . HTTPS is the most obvious example - in the end there are not rules about what certificate root to honor , how to annotate it , etc .. those are decisions implementations make . Yes , but if a client chooses to make bad decisions and trust a DNS server that they have n't ' vetted ' , then are they actually in contravention of this specification ? Actually , I guess it 's really my second point that 's in some ways the most important - typically a MUST / REQUIRED / SHALL specifies an action to take and there will be some obvious consequences for not taking that action ( an error thrown ) . In this case , we 're saying "" You MUST check that the DNS server you want to connect to is trustworthy "" without saying how they should do that . And since there will be no error message from an untrustworthy DNS server , how would they know ? It feels like we 're taking things that outside of the spec ( whether / how to trust a DNS server ) and making it part of the spec ... MUST has many purposes . One of them is to avoid confusion . There 's been a lot of concern during the DOH process that DOH clients will spray queries to random webservers , or otherwise violate security common sense . This "" requirement "" is intended to respond to those concerns , and clarify that we are not proposing to do anything reckless . Well OK then . I 'm still not entirely happy , but if the consensus is to leave it as it is , I 'm not going to push it . I have moved the discussion to the mailing list :" e708a816a1b9653a50b2ec3dac8b2a5108012663 e708a816a1b9653a50b2ec3dac8b2a5108012663 draft-ietf-doh-dns-over-https-latest f830309268e3712db11e75cc782935a39278ca62dd8c819d555be1200be185a0 ['3f9d831263782d3076308a307802b6e144ea35b0735a0d2272d113d667cca662', 'f830309268e3712db11e75cc782935a39278ca62dd8c819d555be1200be185a0'] https://www.ietf.org/mail-archive/web/doh/current/msg00557.html B "in the absence of information about the authenticity of responses, such as dnssec, a dns api server can give a client invalid data in responses. a client must not use arbitrary dns api servers. instead, a client must only use dns api servers specified using mechanisms such as explicit configuration. this does not guarantee protection against invalid data but reduces the risk." 173 {16, 11} {11: 0, 16: 1} {173: 0, 91: 1} "Session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (Section 10.6 of RFC7540) and padding (Section 10.7 of RFC7540)." merge this if you 're ok with the change 9c411d7f9f18341b5f06e04b84d17ee73f8df6e8 9c411d7f9f18341b5f06e04b84d17ee73f8df6e8 draft-ietf-doh-dns-over-https-latest 11057a9419a7e65c6b23f3135e3f7154dce0d6decbddac19d19dafa5b05f4f61 ['11057a9419a7e65c6b23f3135e3f7154dce0d6decbddac19d19dafa5b05f4f61', '62e179c79e18eec9c2f51804fc5434922b2771eedcd695da2301ad0b658f45bf'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/173/commits/19d157d3575085ec50082a80174e8e2570ad5c06 A "session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (section 10.6 of rfc7540) and padding (section 10.7 of rfc7540)." 173 {16, 11} {11: 0, 16: 1} {173: 0, 91: 1} "Session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (Section 10.6 of RFC7540) and padding (Section 10.7 of RFC7540). DNS API Servers can also add DNS padding RFC7830 if the DNS API requests it in the DNS query." merge this if you 're ok with the change 9c411d7f9f18341b5f06e04b84d17ee73f8df6e8 9c411d7f9f18341b5f06e04b84d17ee73f8df6e8 draft-ietf-doh-dns-over-https-latest 62e179c79e18eec9c2f51804fc5434922b2771eedcd695da2301ad0b658f45bf ['11057a9419a7e65c6b23f3135e3f7154dce0d6decbddac19d19dafa5b05f4f61', '62e179c79e18eec9c2f51804fc5434922b2771eedcd695da2301ad0b658f45bf'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/173/commits/19d157d3575085ec50082a80174e8e2570ad5c06 B "session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (section 10.6 of rfc7540) and padding (section 10.7 of rfc7540). dns api servers can also add dns padding rfc7830 if the dns api requests it in the dns query." 172 {100, 114, 24, 27, 28} {27: 0, 24: 1, 28: 1, 114: 2, 100: 3} {172: 0, 157: 1, 174: 1, 176: 2, 178: 3} "In the absence of information about the authenticity of responses, such as DNSSEC, a DNS API server can poison a client's cache. A client MUST NOT authorize arbitrary DNS API servers. Instead, a client MUST specifically authorize DNS API servers using mechanisms such as explicit configuration." This is the only place where a client cache is mentioned , and stub resolvers normally do n't have caches . ( Some browsers have caches for the stub resolvers , but this is rare and not well known outside a small group of people . ) This change replaces that one instance with wording that still explains the main problem . d75960c2c278738b5f56de6fadc484914f2831dd d75960c2c278738b5f56de6fadc484914f2831dd draft-ietf-doh-dns-over-https-latest 30a57b64dda04890e0065af79b6f1b7eb746f4ce2e6557b3f6ecf865457e651d ['3f9d831263782d3076308a307802b6e144ea35b0735a0d2272d113d667cca662', '30a57b64dda04890e0065af79b6f1b7eb746f4ce2e6557b3f6ecf865457e651d'] A "in the absence of information about the authenticity of responses, such as dnssec, a dns api server can poison a client's cache. a client must not authorize arbitrary dns api servers. instead, a client must specifically authorize dns api servers using mechanisms such as explicit configuration." 172 {100, 114, 24, 27, 28} {27: 0, 24: 1, 28: 1, 114: 2, 100: 3} {172: 0, 157: 1, 174: 1, 176: 2, 178: 3} "In the absence of information about the authenticity of responses, such as DNSSEC, a DNS API server can give a client invalid data in responses. A client MUST NOT authorize arbitrary DNS API servers. Instead, a client MUST specifically authorize DNS API servers using mechanisms such as explicit configuration." This is the only place where a client cache is mentioned , and stub resolvers normally do n't have caches . ( Some browsers have caches for the stub resolvers , but this is rare and not well known outside a small group of people . ) This change replaces that one instance with wording that still explains the main problem . d75960c2c278738b5f56de6fadc484914f2831dd d75960c2c278738b5f56de6fadc484914f2831dd draft-ietf-doh-dns-over-https-latest 3f9d831263782d3076308a307802b6e144ea35b0735a0d2272d113d667cca662 ['3f9d831263782d3076308a307802b6e144ea35b0735a0d2272d113d667cca662', '30a57b64dda04890e0065af79b6f1b7eb746f4ce2e6557b3f6ecf865457e651d'] B "in the absence of information about the authenticity of responses, such as dnssec, a dns api server can give a client invalid data in responses. a client must not authorize arbitrary dns api servers. instead, a client must specifically authorize dns api servers using mechanisms such as explicit configuration." 170 {111, 112, 21, 122, 124} {124: 0, 112: 1, 122: 1, 111: 2, 21: 2} {170: 0, 123: 1, 199: 1, 120: 2, 114: 2} "If a DNS API server responds to a DNS API client with a DNS message that has the TC (truncation) bit set in the header, that indicates that the DNS API server was not able to retrieve a full answer for the query and is providing the best answer it could get. This protocol does not require that a DNS API server that cannot get an untruncated answer send back such an answer; it can instead send back an HTTP error to indicate that it cannot give a useful answer." "Attempt to deal with . I 'm not convinced that this is much better than the current text , but it might help others . I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." e5e3fd00a1b00490d06872648de1f3876547eeca e5e3fd00a1b00490d06872648de1f3876547eeca draft-ietf-doh-dns-over-https-latest ff22f0bfecba010f8dfe93c101d4b4893b7189fad99eb0ee2ffa0052b17c386d ['ff22f0bfecba010f8dfe93c101d4b4893b7189fad99eb0ee2ffa0052b17c386d', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a'] A "if a dns api server responds to a dns api client with a dns message that has the tc (truncation) bit set in the header, that indicates that the dns api server was not able to retrieve a full answer for the query and is providing the best answer it could get. this protocol does not require that a dns api server that cannot get an untruncated answer send back such an answer; it can instead send back an http error to indicate that it cannot give a useful answer." 170 {111, 112, 21, 122, 124} {124: 0, 112: 1, 122: 1, 111: 2, 21: 2} {170: 0, 123: 1, 199: 1, 120: 2, 114: 2} "A DNS API server is allowed to answer queries with any valid DNS response. For example, a valid DNS response might have the TC (truncation) bit set in the DNS header to indicate that the server was not able to retrieve a full answer for the query but is providing the best answer it could get. A DNS API server can reply to queries with an HTTP error for queries that it cannot fulfill. In this same example, a DNS API server could use an HTTP error instead of a non- error response that has the TC bit set." "Attempt to deal with . I 'm not convinced that this is much better than the current text , but it might help others . I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." e5e3fd00a1b00490d06872648de1f3876547eeca e5e3fd00a1b00490d06872648de1f3876547eeca draft-ietf-doh-dns-over-https-latest 026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a ['ff22f0bfecba010f8dfe93c101d4b4893b7189fad99eb0ee2ffa0052b17c386d', '026b5acaa65c8e1d5ad447b97640a6013091cf908c2a1218a5751133d25ab90a'] B "a dns api server is allowed to answer queries with any valid dns response. for example, a valid dns response might have the tc (truncation) bit set in the dns header to indicate that the server was not able to retrieve a full answer for the query but is providing the best answer it could get. a dns api server can reply to queries with an http error for queries that it cannot fulfill. in this same example, a dns api server could use an http error instead of a non- error response that has the tc bit set." 169 {32, 101, 102, 103, 40, 44, 45, 31} {45: 0, 102: 1, 103: 1, 101: 2, 40: 2, 44: 3, 32: 3, 31: 4} {169: 0, 144: 1, 199: 1, 142: 2, 174: 2, 128: 3, 157: 3, 63: 4} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""application/dns-message"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4)." "… but keep the requirement for future formats . Attempts to As input to the process , this was extremely valuable , but now that this is in WGLC , we 're done with it . Does the audience of the ultimate RFC need this information ? The only thing that I can see that is worth salvaging is a statement to the effect that is n't explicitly supported . Yes , we still need it . We have not gone to IETF Last Call and , despite what some people think , the IETF gets to change a protocol significantly during IETF Last Call . But even after that , once it is an RFC , others will want to update the RFC . It is important to keep the requirements for them as well . Actually , we probably do n't need it in the RFC as long as the one future requirement ( for formats ) is still there . Please see PR" f836fe64a05706fb0ddf8d809020640e8071e6e0 f836fe64a05706fb0ddf8d809020640e8071e6e0 draft-ietf-doh-dns-over-https-latest 694c964ba47aafa697ebc500c63a9226be552f6381021db6623b1ae2f20f9af9 ['694c964ba47aafa697ebc500c63a9226be552f6381021db6623b1ae2f20f9af9', '31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221'] A "in order to maximize interoperability, dns api clients and dns api servers must support the ""application/dns-message"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4)." 169 {32, 101, 102, 103, 40, 44, 45, 31} {45: 0, 102: 1, 103: 1, 101: 2, 40: 2, 44: 3, 32: 3, 31: 4} {169: 0, 144: 1, 199: 1, 142: 2, 174: 2, 128: 3, 157: 3, 63: 4} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""application/dns-message"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4). Those media types MUST be flexible enough to express every DNS query that would normally be sent in DNS over UDP (including queries and responses that use DNS extensions, but not those that require multiple responses)." "… but keep the requirement for future formats . Attempts to As input to the process , this was extremely valuable , but now that this is in WGLC , we 're done with it . Does the audience of the ultimate RFC need this information ? The only thing that I can see that is worth salvaging is a statement to the effect that is n't explicitly supported . Yes , we still need it . We have not gone to IETF Last Call and , despite what some people think , the IETF gets to change a protocol significantly during IETF Last Call . But even after that , once it is an RFC , others will want to update the RFC . It is important to keep the requirements for them as well . Actually , we probably do n't need it in the RFC as long as the one future requirement ( for formats ) is still there . Please see PR" f836fe64a05706fb0ddf8d809020640e8071e6e0 f836fe64a05706fb0ddf8d809020640e8071e6e0 draft-ietf-doh-dns-over-https-latest 31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221 ['694c964ba47aafa697ebc500c63a9226be552f6381021db6623b1ae2f20f9af9', '31dc49eec35adc06f68ac30030f8e8b2929c2b0ed11a6481463a0734046a9221'] B "in order to maximize interoperability, dns api clients and dns api servers must support the ""application/dns-message"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4). those media types must be flexible enough to express every dns query that would normally be sent in dns over udp (including queries and responses that use dns extensions, but not those that require multiple responses)." 165 {121, 59, 61, 47} {121: 0, 59: 1, 61: 2, 47: 2} {165: 0, 34: 1, 1: 2, 45: 2} The minimum version of HTTP used by DOH SHOULD be HTTP/2 RFC7540. "what motivates this ? Is it just the word minimum ? It changes from normative to non - normative which is a pretty big change and does n't reflect , imo , the consensus we had in singapore . I see s mail now - but I think its in the rough at least with the current text being too strong ( rewording welcome ) . There was considerable discussion about h2 really being necessary to make this work well - there are just practical considerations to being able to always deploy that .. therefore SHOULD ( and the reason you would n't is its impractical to know or enforce version levels across a chain of infrastructure ) . This sounds significant enough to be discussed on the list ." c823b017e0d24aecffca6c46aa8975e0d58ae34f c823b017e0d24aecffca6c46aa8975e0d58ae34f draft-ietf-doh-dns-over-https-latest fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154 ['fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154', '465585c0357edcf53c915ca9703e72cecd063bc3d45cdc8e95c1fb3eb509bb76'] A the minimum version of http used by doh should be http/2 rfc7540. 165 {121, 59, 61, 47} {121: 0, 59: 1, 61: 2, 47: 2} {165: 0, 34: 1, 1: 2, 45: 2} "HTTP/2 RFC7540 is the minimum RECOMMENDED version of HTTP for use with DOH." "what motivates this ? Is it just the word minimum ? It changes from normative to non - normative which is a pretty big change and does n't reflect , imo , the consensus we had in singapore . I see s mail now - but I think its in the rough at least with the current text being too strong ( rewording welcome ) . There was considerable discussion about h2 really being necessary to make this work well - there are just practical considerations to being able to always deploy that .. therefore SHOULD ( and the reason you would n't is its impractical to know or enforce version levels across a chain of infrastructure ) . This sounds significant enough to be discussed on the list ." c823b017e0d24aecffca6c46aa8975e0d58ae34f c823b017e0d24aecffca6c46aa8975e0d58ae34f draft-ietf-doh-dns-over-https-latest 465585c0357edcf53c915ca9703e72cecd063bc3d45cdc8e95c1fb3eb509bb76 ['fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154', '465585c0357edcf53c915ca9703e72cecd063bc3d45cdc8e95c1fb3eb509bb76'] B "http/2 rfc7540 is the minimum recommended version of http for use with doh." 157 {32, 101, 102, 103, 40, 44, 45, 31} {32: 0, 40: 1, 31: 1, 103: 2, 45: 3, 102: 4, 101: 5, 44: 6} {157: 0, 174: 1, 63: 1, 199: 2, 169: 3, 144: 4, 142: 5, 128: 6} "Before using DOH response data for DNS resolution, the client MUST establish that the HTTP request URI is a trusted service for the DOH query. For HTTP requests initiated by the DNS API client this trust is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request. This specification does not extend DNS resolution privileges to URIs that are not recognized by the client as trusted DNS API servers." "This is OK , but I like the changes in better because it brings the "" trust "" discussion further forwards . Preferences ? I prefer this PR over - changing from trust to authorization is more on point I agree authorize is closer to the correct concept here but I 'd like to see a definition of it ( in a section of it 's own early in the document ) . But - I do n't believe there is an existing concept in DNS or DNSSEC that this is re - using ? Does ' authorize ' here mean that : the client will use all response from that server for resolution the client will cache all responses received from that server Also , if the trusted resolver itself is n't doing DNSSEC validation then the client 's cache can still be poisoned . The only 100 % sure method to stop cache poisoning is for the caching entity to perform DNSSEC validation ." 1d0459eea392d50a8c67f56c249876c74ef5122f 1d0459eea392d50a8c67f56c249876c74ef5122f draft-ietf-doh-dns-over-https-latest 48cd3653ab75fd1e0d9d7cc8d08a5c555f4209a8404b6d4ff99167d9650e387e ['48cd3653ab75fd1e0d9d7cc8d08a5c555f4209a8404b6d4ff99167d9650e387e', '61088f578d4ba35f0e73790d7f9ad91298a034e71ed8e0ea5514c5041daf65d4'] A "before using doh response data for dns resolution, the client must establish that the http request uri is a trusted service for the doh query. for http requests initiated by the dns api client this trust is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request. this specification does not extend dns resolution privileges to uris that are not recognized by the client as trusted dns api servers." 157 {32, 101, 102, 103, 40, 44, 45, 31} {32: 0, 40: 1, 31: 1, 103: 2, 45: 3, 102: 4, 101: 5, 44: 6} {157: 0, 174: 1, 63: 1, 199: 2, 169: 3, 144: 4, 142: 5, 128: 6} "Before using DOH response data for DNS resolution, the client MUST establish that the HTTP request URI is authorized for the DOH query. For HTTP requests initiated by the DNS API client this authorization is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request. This specification does not extend DNS resolution privileges to URIs that are not recognized by the client as authorized DNS API servers." "This is OK , but I like the changes in better because it brings the "" trust "" discussion further forwards . Preferences ? I prefer this PR over - changing from trust to authorization is more on point I agree authorize is closer to the correct concept here but I 'd like to see a definition of it ( in a section of it 's own early in the document ) . But - I do n't believe there is an existing concept in DNS or DNSSEC that this is re - using ? Does ' authorize ' here mean that : the client will use all response from that server for resolution the client will cache all responses received from that server Also , if the trusted resolver itself is n't doing DNSSEC validation then the client 's cache can still be poisoned . The only 100 % sure method to stop cache poisoning is for the caching entity to perform DNSSEC validation ." 1d0459eea392d50a8c67f56c249876c74ef5122f 1d0459eea392d50a8c67f56c249876c74ef5122f draft-ietf-doh-dns-over-https-latest 61088f578d4ba35f0e73790d7f9ad91298a034e71ed8e0ea5514c5041daf65d4 ['48cd3653ab75fd1e0d9d7cc8d08a5c555f4209a8404b6d4ff99167d9650e387e', '61088f578d4ba35f0e73790d7f9ad91298a034e71ed8e0ea5514c5041daf65d4'] B "before using doh response data for dns resolution, the client must establish that the http request uri is authorized for the doh query. for http requests initiated by the dns api client this authorization is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request. this specification does not extend dns resolution privileges to uris that are not recognized by the client as authorized dns api servers." 157 {100, 114, 24, 27, 28} {24: 0, 27: 1, 28: 2, 114: 3, 100: 4} {157: 0, 172: 1, 174: 2, 176: 3, 178: 4} "A server that is acting both as a normal web server and a DNS API server is in a position to choose which DNS names it forces a client to resolve (through its web service) and also be the one to answer those queries (through its DNS API service). An untrusted DNS API server can thus easily cause damage by poisoning a client's cache with names that the DNS API server chooses to poison. A client MUST NOT trust a DNS API server simply because it was discovered, or because the client was told to trust the DNS API server by an untrusted party. Instead, a client MUST only trust DNS API server that is configured as trustworthy." "This is OK , but I like the changes in better because it brings the "" trust "" discussion further forwards . Preferences ? I prefer this PR over - changing from trust to authorization is more on point I agree authorize is closer to the correct concept here but I 'd like to see a definition of it ( in a section of it 's own early in the document ) . But - I do n't believe there is an existing concept in DNS or DNSSEC that this is re - using ? Does ' authorize ' here mean that : the client will use all response from that server for resolution the client will cache all responses received from that server Also , if the trusted resolver itself is n't doing DNSSEC validation then the client 's cache can still be poisoned . The only 100 % sure method to stop cache poisoning is for the caching entity to perform DNSSEC validation ." 1d0459eea392d50a8c67f56c249876c74ef5122f 1d0459eea392d50a8c67f56c249876c74ef5122f draft-ietf-doh-dns-over-https-latest 33e4d1bc794802d0469a6e9ba8246aa96f1813e31e1ef10baee2c534698b50c4 ['33e4d1bc794802d0469a6e9ba8246aa96f1813e31e1ef10baee2c534698b50c4', '30a57b64dda04890e0065af79b6f1b7eb746f4ce2e6557b3f6ecf865457e651d'] A "a server that is acting both as a normal web server and a dns api server is in a position to choose which dns names it forces a client to resolve (through its web service) and also be the one to answer those queries (through its dns api service). an untrusted dns api server can thus easily cause damage by poisoning a client's cache with names that the dns api server chooses to poison. a client must not trust a dns api server simply because it was discovered, or because the client was told to trust the dns api server by an untrusted party. instead, a client must only trust dns api server that is configured as trustworthy." 157 {100, 114, 24, 27, 28} {24: 0, 27: 1, 28: 2, 114: 3, 100: 4} {157: 0, 172: 1, 174: 2, 176: 3, 178: 4} "In the absence of information about the authenticity of responses, such as DNSSEC, a DNS API server can poison a client's cache. A client MUST NOT authorize arbitrary DNS API servers. Instead, a client MUST specifically authorize DNS API servers using mechanisms such as explicit configuration." "This is OK , but I like the changes in better because it brings the "" trust "" discussion further forwards . Preferences ? I prefer this PR over - changing from trust to authorization is more on point I agree authorize is closer to the correct concept here but I 'd like to see a definition of it ( in a section of it 's own early in the document ) . But - I do n't believe there is an existing concept in DNS or DNSSEC that this is re - using ? Does ' authorize ' here mean that : the client will use all response from that server for resolution the client will cache all responses received from that server Also , if the trusted resolver itself is n't doing DNSSEC validation then the client 's cache can still be poisoned . The only 100 % sure method to stop cache poisoning is for the caching entity to perform DNSSEC validation ." 1d0459eea392d50a8c67f56c249876c74ef5122f 1d0459eea392d50a8c67f56c249876c74ef5122f draft-ietf-doh-dns-over-https-latest 30a57b64dda04890e0065af79b6f1b7eb746f4ce2e6557b3f6ecf865457e651d ['33e4d1bc794802d0469a6e9ba8246aa96f1813e31e1ef10baee2c534698b50c4', '30a57b64dda04890e0065af79b6f1b7eb746f4ce2e6557b3f6ecf865457e651d'] B "in the absence of information about the authenticity of responses, such as dnssec, a dns api server can poison a client's cache. a client must not authorize arbitrary dns api servers. instead, a client must specifically authorize dns api servers using mechanisms such as explicit configuration." 154 {0, 1, 74, 43, 62} {62: 0, 1: 1, 0: 2, 74: 2, 43: 2} {154: 0, 177: 1, 199: 2, 56: 2} "A DOH response that was previously stored in an HTTP cache will contain the RFC7234 Age response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response. DNS API clients should subtract this time from the DNS TTL if they are re-sharing the information in a non HTTP context (e.g., their own DNS cache) to determine the remaining time to live of the DNS record." unnecessary . The point here is to highlight the need to tweak the TTL to account for Age , not to go into why . It 's actually a generic requirement that is not specific to translation to other contexts . 34595b01791c2e29b47a075249d0e58643391311 34595b01791c2e29b47a075249d0e58643391311 draft-ietf-doh-dns-over-https-latest 87d516551d51b3d1ebc937919ba39edb952de3b91acaa6f9850e320ec978df34 ['87d516551d51b3d1ebc937919ba39edb952de3b91acaa6f9850e320ec978df34', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074'] A "a doh response that was previously stored in an http cache will contain the rfc7234 age response header indicating the elapsed time between when the entry was placed in the http cache and the current doh response. dns api clients should subtract this time from the dns ttl if they are re-sharing the information in a non http context (e.g., their own dns cache) to determine the remaining time to live of the dns record." 154 {0, 1, 74, 43, 62} {62: 0, 1: 1, 0: 2, 74: 2, 43: 2} {154: 0, 177: 1, 199: 2, 56: 2} "A DOH response that was previously stored in an HTTP cache will contain the RFC7234 Age response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response. DNS API clients should subtract this time from the DNS TTL in each RRset to determine the time remaining on the TTL." unnecessary . The point here is to highlight the need to tweak the TTL to account for Age , not to go into why . It 's actually a generic requirement that is not specific to translation to other contexts . 34595b01791c2e29b47a075249d0e58643391311 34595b01791c2e29b47a075249d0e58643391311 draft-ietf-doh-dns-over-https-latest a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074 ['87d516551d51b3d1ebc937919ba39edb952de3b91acaa6f9850e320ec978df34', 'a3262f1f4650d5c2b8358dd4c71bbdcf5b16a4c53208d88a1893a340f9ec2074'] B "a doh response that was previously stored in an http cache will contain the rfc7234 age response header indicating the elapsed time between when the entry was placed in the http cache and the current doh response. dns api clients should subtract this time from the dns ttl in each rrset to determine the time remaining on the ttl." 147 {82} {82: 0} {147: 0} "A server that supports this protocol on one or more URIs is called a ""DNS API server"" to differentiate it from a ""DNS server"" (one that uses the regular DNS protocol). Similarly, a client that supports this protocol is called a ""DNS API client""." "This is n't necessary . I realize that this reverts a deliberate change as part of , but if it is important to describe - which I do n't think helps - then I do n't think that this is the right approach . Adding an explicit statement would be better . hmm .. I think this was committed a little prematurely .. as says , the change was originally deliberate to emphasize that uris are the primitive and not the server . Happy to make that an explicit statement - but do n't want to leave it out ." 74f1bb071f2e566ba1e69b2a375e47643e153d06 74f1bb071f2e566ba1e69b2a375e47643e153d06 draft-ietf-doh-dns-over-https-latest b66d70acdebb9d8625eb60ce1dcef448418bee2f031f76440b937dbc6db7f80d ['b66d70acdebb9d8625eb60ce1dcef448418bee2f031f76440b937dbc6db7f80d', '0f999c14b18da6253b56925e3142fdf439ac883d1f734c1bc2b267198c35a011'] A "a server that supports this protocol on one or more uris is called a ""dns api server"" to differentiate it from a ""dns server"" (one that uses the regular dns protocol). similarly, a client that supports this protocol is called a ""dns api client""." 147 {82} {82: 0} {147: 0} "A server that supports this protocol is called a ""DNS API server"" to differentiate it from a ""DNS server"" (one that uses the regular DNS protocol). Similarly, a client that supports this protocol is called a ""DNS API client""." "This is n't necessary . I realize that this reverts a deliberate change as part of , but if it is important to describe - which I do n't think helps - then I do n't think that this is the right approach . Adding an explicit statement would be better . hmm .. I think this was committed a little prematurely .. as says , the change was originally deliberate to emphasize that uris are the primitive and not the server . Happy to make that an explicit statement - but do n't want to leave it out ." 74f1bb071f2e566ba1e69b2a375e47643e153d06 74f1bb071f2e566ba1e69b2a375e47643e153d06 draft-ietf-doh-dns-over-https-latest 0f999c14b18da6253b56925e3142fdf439ac883d1f734c1bc2b267198c35a011 ['b66d70acdebb9d8625eb60ce1dcef448418bee2f031f76440b937dbc6db7f80d', '0f999c14b18da6253b56925e3142fdf439ac883d1f734c1bc2b267198c35a011'] B "a server that supports this protocol is called a ""dns api server"" to differentiate it from a ""dns server"" (one that uses the regular dns protocol). similarly, a client that supports this protocol is called a ""dns api client""." 146 {80, 115} {80: 0, 115: 1} {146: 0, 76: 1} "Two primary uses cases were considered during this protocol's development. They included preventing on-path devices from interfering with DNS operations and allowing web applications to access DNS information via existing browser APIs in a safe way consistent with Cross Origin Resource Sharing (CORS) CORS. There are certainly other uses for this work." "When a statement is so definite , the reader tends to assume that there is something more to be said . In this case , the obvious question is "" what other uses are there ? "" , closely followed by "" and what reason could the authors have had for not mentioning them ? """ 507de935837a80f35befbdd60019a6b16612fac6 507de935837a80f35befbdd60019a6b16612fac6 draft-ietf-doh-dns-over-https-latest f8a2280983a1e319e157524334815a3adc7354805fa7e801feb499216bf9e2c6 ['b0874eb1b634f28373a0bf7aebdcdda0478b9515db476aeda2ae540f9afae9a7', 'f8a2280983a1e319e157524334815a3adc7354805fa7e801feb499216bf9e2c6'] A "two primary uses cases were considered during this protocol's development. they included preventing on-path devices from interfering with dns operations and allowing web applications to access dns information via existing browser apis in a safe way consistent with cross origin resource sharing (cors) cors. there are certainly other uses for this work." 146 {80, 115} {80: 0, 115: 1} {146: 0, 76: 1} "Two primary uses cases were considered during this protocol's development. They included preventing on-path devices from interfering with DNS operations and allowing web applications to access DNS information via existing browser APIs in a safe way consistent with Cross Origin Resource Sharing (CORS) CORS. No special effort has been taken to enable or prevent application to other use cases." "When a statement is so definite , the reader tends to assume that there is something more to be said . In this case , the obvious question is "" what other uses are there ? "" , closely followed by "" and what reason could the authors have had for not mentioning them ? """ 507de935837a80f35befbdd60019a6b16612fac6 507de935837a80f35befbdd60019a6b16612fac6 draft-ietf-doh-dns-over-https-latest b0874eb1b634f28373a0bf7aebdcdda0478b9515db476aeda2ae540f9afae9a7 ['b0874eb1b634f28373a0bf7aebdcdda0478b9515db476aeda2ae540f9afae9a7', 'f8a2280983a1e319e157524334815a3adc7354805fa7e801feb499216bf9e2c6'] B "two primary uses cases were considered during this protocol's development. they included preventing on-path devices from interfering with dns operations and allowing web applications to access dns information via existing browser apis in a safe way consistent with cross origin resource sharing (cors) cors. no special effort has been taken to enable or prevent application to other use cases." 145 {} {} {} "This document describes how to run DNS service over HTTP (DOH) using https:// URIs." "The "" using https:// URIs "" part is just odd . Easier to simply state that this uses HTTPS at this level . Also , if the concept of a DNS service evolves to encompass other types of exchange than simple request - response , then "" run DNS service "" wo n't be accurate . I "" DOH "" to the title because it was now being removed from the Abstract ." 5fbfa99b817d2d4e2aef2c22075d51297520c727 5fbfa99b817d2d4e2aef2c22075d51297520c727 draft-ietf-doh-dns-over-https-latest 544d6340f6b637a117d7b6660061af671f3773de33b7240861b48e66d73ac3e6 ['9d190ccbbad8009eef8cd59f2ccde0389155de4daa5ab68073df85bf708fca81', '628da26fcdc51f0bf32ab2472d74da82447d9350ab9a0fb5c71bdeb25e2d39fd', '544d6340f6b637a117d7b6660061af671f3773de33b7240861b48e66d73ac3e6', '265cb089136a19eb72c0515eac7805081bda999a3c52413dc8dd2f995df33651'] A "this document describes how to run dns service over http (doh) using https:// uris." 145 {} {} {} "This document defines a specific protocol for sending DNS RFC1035 queries and getting DNS responses over HTTP RFC7540 using https:// (and therefore TLS RFC5246 security for integrity and confidentiality). Each DNS query-response pair is mapped into a HTTP exchange." "The "" using https:// URIs "" part is just odd . Easier to simply state that this uses HTTPS at this level . Also , if the concept of a DNS service evolves to encompass other types of exchange than simple request - response , then "" run DNS service "" wo n't be accurate . I "" DOH "" to the title because it was now being removed from the Abstract ." 5fbfa99b817d2d4e2aef2c22075d51297520c727 5fbfa99b817d2d4e2aef2c22075d51297520c727 draft-ietf-doh-dns-over-https-latest 265cb089136a19eb72c0515eac7805081bda999a3c52413dc8dd2f995df33651 ['9d190ccbbad8009eef8cd59f2ccde0389155de4daa5ab68073df85bf708fca81', '628da26fcdc51f0bf32ab2472d74da82447d9350ab9a0fb5c71bdeb25e2d39fd', '544d6340f6b637a117d7b6660061af671f3773de33b7240861b48e66d73ac3e6', '265cb089136a19eb72c0515eac7805081bda999a3c52413dc8dd2f995df33651'] A "this document defines a specific protocol for sending dns rfc1035 queries and getting dns responses over http rfc7540 using https:// (and therefore tls rfc5246 security for integrity and confidentiality). each dns query-response pair is mapped into a http exchange." 145 {} {} {} This document describes how to make DNS queries over HTTPS. "The "" using https:// URIs "" part is just odd . Easier to simply state that this uses HTTPS at this level . Also , if the concept of a DNS service evolves to encompass other types of exchange than simple request - response , then "" run DNS service "" wo n't be accurate . I "" DOH "" to the title because it was now being removed from the Abstract ." 5fbfa99b817d2d4e2aef2c22075d51297520c727 5fbfa99b817d2d4e2aef2c22075d51297520c727 draft-ietf-doh-dns-over-https-latest 628da26fcdc51f0bf32ab2472d74da82447d9350ab9a0fb5c71bdeb25e2d39fd ['9d190ccbbad8009eef8cd59f2ccde0389155de4daa5ab68073df85bf708fca81', '628da26fcdc51f0bf32ab2472d74da82447d9350ab9a0fb5c71bdeb25e2d39fd', '544d6340f6b637a117d7b6660061af671f3773de33b7240861b48e66d73ac3e6', '265cb089136a19eb72c0515eac7805081bda999a3c52413dc8dd2f995df33651'] B this document describes how to make dns queries over https. 145 {} {} {} "This document defines a specific protocol for sending DNS RFC1035 queries and getting DNS responses over HTTP RFC7540 using https URIs (and therefore TLS RFC5246 security for integrity and confidentiality). Each DNS query-response pair is mapped into a HTTP exchange." "The "" using https:// URIs "" part is just odd . Easier to simply state that this uses HTTPS at this level . Also , if the concept of a DNS service evolves to encompass other types of exchange than simple request - response , then "" run DNS service "" wo n't be accurate . I "" DOH "" to the title because it was now being removed from the Abstract ." 5fbfa99b817d2d4e2aef2c22075d51297520c727 5fbfa99b817d2d4e2aef2c22075d51297520c727 draft-ietf-doh-dns-over-https-latest 9d190ccbbad8009eef8cd59f2ccde0389155de4daa5ab68073df85bf708fca81 ['9d190ccbbad8009eef8cd59f2ccde0389155de4daa5ab68073df85bf708fca81', '628da26fcdc51f0bf32ab2472d74da82447d9350ab9a0fb5c71bdeb25e2d39fd', '544d6340f6b637a117d7b6660061af671f3773de33b7240861b48e66d73ac3e6', '265cb089136a19eb72c0515eac7805081bda999a3c52413dc8dd2f995df33651'] B "this document defines a specific protocol for sending dns rfc1035 queries and getting dns responses over http rfc7540 using https uris (and therefore tls rfc5246 security for integrity and confidentiality). each dns query-response pair is mapped into a http exchange." 144 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {107: 0, 106: 1, 53: 1, 54: 2, 5: 3, 64: 4, 4: 4, 12: 5, 98: 5, 99: 6} {144: 0, 142: 1, 199: 1, 131: 2, 65: 3, 61: 4, 78: 4, 20: 5} "The DNS API client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""message/dns"" (as described in dnswire) responses but MAY also process any other type it receives." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] A "the dns api client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""message/dns"" (as described in dnswire) responses but may also process any other type it receives." 144 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {107: 0, 106: 1, 53: 1, 54: 2, 5: 3, 64: 4, 4: 4, 12: 5, 98: 5, 99: 6} {144: 0, 142: 1, 199: 1, 131: 2, 65: 3, 61: 4, 78: 4, 20: 5} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as message/dns, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as message/ dns. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] A "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as message/dns, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as message/ dns. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 144 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {107: 0, 106: 1, 53: 1, 54: 2, 5: 3, 64: 4, 4: 4, 12: 5, 98: 5, 99: 6} {144: 0, 142: 1, 199: 1, 131: 2, 65: 3, 61: 4, 78: 4, 20: 5} "The DNS API client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""application/dns-message"" (as described in dnswire) responses but MAY also process any other type it receives." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] B "the dns api client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""application/dns-message"" (as described in dnswire) responses but may also process any other type it receives." 144 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {107: 0, 106: 1, 53: 1, 54: 2, 5: 3, 64: 4, 4: 4, 12: 5, 98: 5, 99: 6} {144: 0, 142: 1, 199: 1, 131: 2, 65: 3, 61: 4, 78: 4, 20: 5} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as application/dns-message, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as application/dns-message. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest d0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', 'd0e91bba5cab2a8ad1ca50beeb6fe06dea2361b844bbe802b0f0849ae76f7e9c', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', '81fb5351bfb05e836fea87c6a634ab4eeb89448b3b0adabf319f385d05215fa4'] B "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as application/dns-message, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as application/dns-message. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 144 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {99: 0, 98: 1, 4: 2, 5: 3, 64: 4, 54: 4, 12: 5, 53: 5, 107: 6, 106: 7} {144: 0, 142: 1, 78: 2, 65: 3, 61: 4, 131: 4, 20: 5, 199: 5} The requests are represented as message/dns typed bodies. "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest ce0553c881b7126589d43db5a8be4e4604b92869026921b4758076f923a97d48 ['ce0553c881b7126589d43db5a8be4e4604b92869026921b4758076f923a97d48', 'aaa6f87f1b6c314d4a7a736c8a602d9dce8e0e6386b8bb3344582a0fa61d5672'] A the requests are represented as message/dns typed bodies. 144 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {99: 0, 98: 1, 4: 2, 5: 3, 64: 4, 54: 4, 12: 5, 53: 5, 107: 6, 106: 7} {144: 0, 142: 1, 78: 2, 65: 3, 61: 4, 131: 4, 20: 5, 199: 5} The requests are represented as application/dns-message typed bodies. "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest aaa6f87f1b6c314d4a7a736c8a602d9dce8e0e6386b8bb3344582a0fa61d5672 ['ce0553c881b7126589d43db5a8be4e4604b92869026921b4758076f923a97d48', 'aaa6f87f1b6c314d4a7a736c8a602d9dce8e0e6386b8bb3344582a0fa61d5672'] B the requests are represented as application/dns-message typed bodies. 144 {38, 70, 19, 20, 118, 94} {70: 0, 118: 1, 19: 2, 38: 2, 94: 3, 20: 4} {144: 0, 142: 1, 78: 2, 122: 2, 43: 3, 1: 4} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""message/dns"", but it is possible that other response formats will be defined in the future." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799 ['901ffd2f36787c6dde79bcd401f9e2404def3aa0a84fd3ee4df9b73098079416', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', 'c20cf1aa0c63b34466c380e885bf9e6324890648b38cb339dad9f71379aa722b', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] A "at the time this is published, the response types are works in progress. the only response type defined in this document is ""message/dns"", but it is possible that other response formats will be defined in the future." 144 {38, 70, 19, 20, 118, 94} {70: 0, 118: 1, 19: 2, 38: 2, 94: 3, 20: 4} {144: 0, 142: 1, 78: 2, 122: 2, 43: 3, 1: 4} "The DNS response for ""message/dns"" in dnswire MAY have one or more EDNS options, depending on the extension definition of the extensions given in the DNS request." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa ['901ffd2f36787c6dde79bcd401f9e2404def3aa0a84fd3ee4df9b73098079416', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', 'c20cf1aa0c63b34466c380e885bf9e6324890648b38cb339dad9f71379aa722b', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] A "the dns response for ""message/dns"" in dnswire may have one or more edns options, depending on the extension definition of the extensions given in the dns request." 144 {38, 70, 19, 20, 118, 94} {70: 0, 118: 1, 19: 2, 38: 2, 94: 3, 20: 4} {144: 0, 142: 1, 78: 2, 122: 2, 43: 3, 1: 4} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""application/dns-message"", but it is possible that other response formats will be defined in the future." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 901ffd2f36787c6dde79bcd401f9e2404def3aa0a84fd3ee4df9b73098079416 ['901ffd2f36787c6dde79bcd401f9e2404def3aa0a84fd3ee4df9b73098079416', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', 'c20cf1aa0c63b34466c380e885bf9e6324890648b38cb339dad9f71379aa722b', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] B "at the time this is published, the response types are works in progress. the only response type defined in this document is ""application/dns-message"", but it is possible that other response formats will be defined in the future." 144 {38, 70, 19, 20, 118, 94} {70: 0, 118: 1, 19: 2, 38: 2, 94: 3, 20: 4} {144: 0, 142: 1, 78: 2, 122: 2, 43: 3, 1: 4} "The DNS response for ""application/dns-message"" in dnswire MAY have one or more EDNS options, depending on the extension definition of the extensions given in the DNS request." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest c20cf1aa0c63b34466c380e885bf9e6324890648b38cb339dad9f71379aa722b ['901ffd2f36787c6dde79bcd401f9e2404def3aa0a84fd3ee4df9b73098079416', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', 'c20cf1aa0c63b34466c380e885bf9e6324890648b38cb339dad9f71379aa722b', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] B "the dns response for ""application/dns-message"" in dnswire may have one or more edns options, depending on the extension definition of the extensions given in the dns request." 144 {50, 51} {50: 0, 51: 1} {144: 0, 142: 1} "A DNS API server MUST be able to process message/dns request messages." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 7414e2891748fd01d0005479179c9c6a3434fc2e4c6961e22b00738e26391af0 ['7414e2891748fd01d0005479179c9c6a3434fc2e4c6961e22b00738e26391af0', '990e6e7a38971284745664168b1595eaf62fc7b0fe39773b9b830a8f13801043'] A "a dns api server must be able to process message/dns request messages." 144 {50, 51} {50: 0, 51: 1} {144: 0, 142: 1} "A DNS API server MUST be able to process application/dns-message request messages." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 990e6e7a38971284745664168b1595eaf62fc7b0fe39773b9b830a8f13801043 ['7414e2891748fd01d0005479179c9c6a3434fc2e4c6961e22b00738e26391af0', '990e6e7a38971284745664168b1595eaf62fc7b0fe39773b9b830a8f13801043'] B "a dns api server must be able to process application/dns-message request messages." 144 {32, 101, 102, 103, 40, 44, 45, 31} {102: 0, 101: 1, 45: 1, 44: 2, 103: 2, 40: 3, 32: 4, 31: 5} {144: 0, 142: 1, 169: 1, 128: 2, 199: 2, 174: 3, 157: 4, 63: 5} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""message/dns"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4)." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest d0b7795392109a5c00ec57dc868086e21a6065923a6e4b0dd309a0f748080ac7 ['d0b7795392109a5c00ec57dc868086e21a6065923a6e4b0dd309a0f748080ac7', '694c964ba47aafa697ebc500c63a9226be552f6381021db6623b1ae2f20f9af9'] A "in order to maximize interoperability, dns api clients and dns api servers must support the ""message/dns"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4)." 144 {32, 101, 102, 103, 40, 44, 45, 31} {102: 0, 101: 1, 45: 1, 44: 2, 103: 2, 40: 3, 32: 4, 31: 5} {144: 0, 142: 1, 169: 1, 128: 2, 199: 2, 174: 3, 157: 4, 63: 5} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""application/dns-message"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4)." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 694c964ba47aafa697ebc500c63a9226be552f6381021db6623b1ae2f20f9af9 ['d0b7795392109a5c00ec57dc868086e21a6065923a6e4b0dd309a0f748080ac7', '694c964ba47aafa697ebc500c63a9226be552f6381021db6623b1ae2f20f9af9'] B "in order to maximize interoperability, dns api clients and dns api servers must support the ""application/dns-message"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4)." 144 {71, 72, 76, 77, 48, 55} {72: 0, 76: 1, 71: 1, 48: 2, 77: 2, 55: 2} {144: 0, 196: 1, 142: 1, 131: 2, 122: 2, 199: 2} "The media type is ""message/dns""." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest 906164d731ba8c07cf16bce19a1a558e9934d924b1e1e8f0fd9fcfe1177cf035 ['906164d731ba8c07cf16bce19a1a558e9934d924b1e1e8f0fd9fcfe1177cf035', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347'] A "the media type is ""message/dns""." 144 {71, 72, 76, 77, 48, 55} {72: 0, 76: 1, 71: 1, 48: 2, 77: 2, 55: 2} {144: 0, 196: 1, 142: 1, 131: 2, 122: 2, 199: 2} "The media type is ""application/dns-message""." "Fixes issue The message/ tree requires that the sub - types contain other MIME messages , which DOH does not . Maybe use application / dns - message instead ." c1262572551171535c90c39f8b032c595649f49d c1262572551171535c90c39f8b032c595649f49d draft-ietf-doh-dns-over-https-latest a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347 ['906164d731ba8c07cf16bce19a1a558e9934d924b1e1e8f0fd9fcfe1177cf035', 'a6c775315db09268f1902c47dda532c32c111963985e4bd07c9b28c67f6ce347'] B "the media type is ""application/dns-message""." 142 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {106: 0, 107: 1, 53: 2, 54: 3, 5: 4, 64: 5, 4: 5, 12: 6, 98: 6, 99: 7} {142: 0, 144: 1, 199: 2, 131: 3, 65: 4, 61: 5, 78: 5, 20: 6} "The DNS API client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""application/dns-udpwireformat"" (See dnswire) responses but MAY also process any other type it receives." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 7d6e41253928c83a898ef7dc82c3f6b032eb5010d247b6521029c6a51b6b05b3 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', 'f1baeb202918f316a23a3dc361f42589e79c215b43f38a0d9b3e80a1172b4738', '7d6e41253928c83a898ef7dc82c3f6b032eb5010d247b6521029c6a51b6b05b3'] A "the dns api client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""application/dns-udpwireformat"" (see dnswire) responses but may also process any other type it receives." 142 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {106: 0, 107: 1, 53: 2, 54: 3, 5: 4, 64: 5, 4: 5, 12: 6, 98: 6, 99: 7} {142: 0, 144: 1, 199: 2, 131: 3, 65: 4, 61: 5, 78: 5, 20: 6} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as application/dns-udpwireformat, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as application/dns-udpwireformat. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest f1baeb202918f316a23a3dc361f42589e79c215b43f38a0d9b3e80a1172b4738 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', 'f1baeb202918f316a23a3dc361f42589e79c215b43f38a0d9b3e80a1172b4738', '7d6e41253928c83a898ef7dc82c3f6b032eb5010d247b6521029c6a51b6b05b3'] A "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as application/dns-udpwireformat, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as application/dns-udpwireformat. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 142 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {106: 0, 107: 1, 53: 2, 54: 3, 5: 4, 64: 5, 4: 5, 12: 6, 98: 6, 99: 7} {142: 0, 144: 1, 199: 2, 131: 3, 65: 4, 61: 5, 78: 5, 20: 6} "The DNS API client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""message/dns"" (as described in dnswire) responses but MAY also process any other type it receives." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', 'f1baeb202918f316a23a3dc361f42589e79c215b43f38a0d9b3e80a1172b4738', '7d6e41253928c83a898ef7dc82c3f6b032eb5010d247b6521029c6a51b6b05b3'] B "the dns api client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""message/dns"" (as described in dnswire) responses but may also process any other type it receives." 142 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {106: 0, 107: 1, 53: 2, 54: 3, 5: 4, 64: 5, 4: 5, 12: 6, 98: 6, 99: 7} {142: 0, 144: 1, 199: 2, 131: 3, 65: 4, 61: 5, 78: 5, 20: 6} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as message/dns, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates the request and response, thus eliminating the need for the ID in a media type such as message/ dns. The use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26 ['d93182ec2bf082537a4999dd7419ae19a0a7b6e8f500eb0828f7a4bb516cba26', '85becb192f23570322e244246740aff361c166679d726cdb40574668408885f5', 'f1baeb202918f316a23a3dc361f42589e79c215b43f38a0d9b3e80a1172b4738', '7d6e41253928c83a898ef7dc82c3f6b032eb5010d247b6521029c6a51b6b05b3'] B "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as message/dns, should use a dns id of 0 in every dns request. http correlates the request and response, thus eliminating the need for the id in a media type such as message/ dns. the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 142 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {98: 0, 99: 1, 4: 1, 5: 2, 64: 3, 54: 3, 12: 4, 53: 4, 107: 5, 106: 6} {142: 0, 144: 1, 78: 1, 65: 2, 61: 3, 131: 3, 20: 4, 199: 4} "The requests are represented as application/dns-udpwirefomat typed bodies." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 021c92cc72fb7bc7a922307a0c01249da467139a386e780da9116a4780bae75f ['ce0553c881b7126589d43db5a8be4e4604b92869026921b4758076f923a97d48', '021c92cc72fb7bc7a922307a0c01249da467139a386e780da9116a4780bae75f'] A "the requests are represented as application/dns-udpwirefomat typed bodies." 142 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {98: 0, 99: 1, 4: 1, 5: 2, 64: 3, 54: 3, 12: 4, 53: 4, 107: 5, 106: 6} {142: 0, 144: 1, 78: 1, 65: 2, 61: 3, 131: 3, 20: 4, 199: 4} The requests are represented as message/dns typed bodies. The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest ce0553c881b7126589d43db5a8be4e4604b92869026921b4758076f923a97d48 ['ce0553c881b7126589d43db5a8be4e4604b92869026921b4758076f923a97d48', '021c92cc72fb7bc7a922307a0c01249da467139a386e780da9116a4780bae75f'] B the requests are represented as message/dns typed bodies. 142 {38, 70, 19, 20, 118, 94} {118: 0, 70: 1, 19: 1, 38: 1, 94: 2, 20: 3} {142: 0, 144: 1, 78: 1, 122: 1, 43: 2, 1: 3} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that other response formats will be defined in the future." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a ['fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', '27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] A "at the time this is published, the response types are works in progress. the only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that other response formats will be defined in the future." 142 {38, 70, 19, 20, 118, 94} {118: 0, 70: 1, 19: 1, 38: 1, 94: 2, 20: 3} {142: 0, 144: 1, 78: 1, 122: 1, 43: 2, 1: 3} "The DNS response for ""application/dns-udpwireformat"" in dnswire MAY have one or more EDNS options, depending on the extension definition of the extensions given in the DNS request." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833 ['fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', '27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] A "the dns response for ""application/dns-udpwireformat"" in dnswire may have one or more edns options, depending on the extension definition of the extensions given in the dns request." 142 {38, 70, 19, 20, 118, 94} {118: 0, 70: 1, 19: 1, 38: 1, 94: 2, 20: 3} {142: 0, 144: 1, 78: 1, 122: 1, 43: 2, 1: 3} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""message/dns"", but it is possible that other response formats will be defined in the future." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799 ['fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', '27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] B "at the time this is published, the response types are works in progress. the only response type defined in this document is ""message/dns"", but it is possible that other response formats will be defined in the future." 142 {38, 70, 19, 20, 118, 94} {118: 0, 70: 1, 19: 1, 38: 1, 94: 2, 20: 3} {142: 0, 144: 1, 78: 1, 122: 1, 43: 2, 1: 3} "The DNS response for ""message/dns"" in dnswire MAY have one or more EDNS options, depending on the extension definition of the extensions given in the DNS request." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa ['fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833', '451ae7afaef81964aba6c1d03753f4bc0e7d0bee9bafae9ffe55b59e4ddd4afa', '27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a', '1728c9c40b975330b8a660e36e88544f4c37f3bff0846b893ef3c747a9a53799'] B "the dns response for ""message/dns"" in dnswire may have one or more edns options, depending on the extension definition of the extensions given in the dns request." 142 {50, 51} {51: 0, 50: 1} {142: 0, 144: 1} "A DNS API server MUST be able to process application/dns- udpwireformat request messages." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest c205a9581877fb669428374690201a87e5e88e1e17fa2e78a14148c62b51c989 ['7414e2891748fd01d0005479179c9c6a3434fc2e4c6961e22b00738e26391af0', 'c205a9581877fb669428374690201a87e5e88e1e17fa2e78a14148c62b51c989'] A "a dns api server must be able to process application/dns- udpwireformat request messages." 142 {50, 51} {51: 0, 50: 1} {142: 0, 144: 1} "A DNS API server MUST be able to process message/dns request messages." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 7414e2891748fd01d0005479179c9c6a3434fc2e4c6961e22b00738e26391af0 ['7414e2891748fd01d0005479179c9c6a3434fc2e4c6961e22b00738e26391af0', 'c205a9581877fb669428374690201a87e5e88e1e17fa2e78a14148c62b51c989'] B "a dns api server must be able to process message/dns request messages." 142 {32, 101, 102, 103, 40, 44, 45, 31} {101: 0, 44: 1, 102: 1, 45: 2, 103: 3, 40: 4, 32: 5, 31: 6} {142: 0, 128: 1, 144: 1, 169: 2, 199: 3, 174: 4, 157: 5, 63: 6} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""application/dns-udpwireformat"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4)." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 691787dee56f6f17a1e77e896482dc026da67a95929e58d3de1237f7274a2325 ['d0b7795392109a5c00ec57dc868086e21a6065923a6e4b0dd309a0f748080ac7', '691787dee56f6f17a1e77e896482dc026da67a95929e58d3de1237f7274a2325'] A "in order to maximize interoperability, dns api clients and dns api servers must support the ""application/dns-udpwireformat"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4)." 142 {32, 101, 102, 103, 40, 44, 45, 31} {101: 0, 44: 1, 102: 1, 45: 2, 103: 3, 40: 4, 32: 5, 31: 6} {142: 0, 128: 1, 144: 1, 169: 2, 199: 3, 174: 4, 157: 5, 63: 6} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""message/dns"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4)." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest d0b7795392109a5c00ec57dc868086e21a6065923a6e4b0dd309a0f748080ac7 ['d0b7795392109a5c00ec57dc868086e21a6065923a6e4b0dd309a0f748080ac7', '691787dee56f6f17a1e77e896482dc026da67a95929e58d3de1237f7274a2325'] B "in order to maximize interoperability, dns api clients and dns api servers must support the ""message/dns"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4)." 142 {71, 72, 76, 77, 48, 55} {71: 0, 72: 1, 76: 2, 48: 3, 77: 3, 55: 3} {142: 0, 144: 1, 196: 2, 131: 3, 122: 3, 199: 3} "The media type is ""application/dns-udpwireformat""." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 50f4cce57486d130ff38bddfb2c668ac840c94413902899262391261fd08f99f ['906164d731ba8c07cf16bce19a1a558e9934d924b1e1e8f0fd9fcfe1177cf035', '50f4cce57486d130ff38bddfb2c668ac840c94413902899262391261fd08f99f'] A "the media type is ""application/dns-udpwireformat""." 142 {71, 72, 76, 77, 48, 55} {71: 0, 72: 1, 76: 2, 48: 3, 77: 3, 55: 3} {142: 0, 144: 1, 196: 2, 131: 3, 122: 3, 199: 3} "The media type is ""message/dns""." The media type has been the subject of some debate . The above might be the best I 've seen . 4b570b8940588ef2593906eee8a8a479468f6ff9 4b570b8940588ef2593906eee8a8a479468f6ff9 draft-ietf-doh-dns-over-https-latest 906164d731ba8c07cf16bce19a1a558e9934d924b1e1e8f0fd9fcfe1177cf035 ['906164d731ba8c07cf16bce19a1a558e9934d924b1e1e8f0fd9fcfe1177cf035', '50f4cce57486d130ff38bddfb2c668ac840c94413902899262391261fd08f99f'] B "the media type is ""message/dns""." 140 {} {} {} "This document does not change the definition of any HTTP response codes or otherwise proscribe their use." the removed text is inconsistent with the immediately preceding sentence . It can simply be removed . e22017fbab657cf0a1d66c11c0f0c584e04b092a e22017fbab657cf0a1d66c11c0f0c584e04b092a draft-ietf-doh-dns-over-https-latest 66bb83086d8506219ac25610bfd9f24c58f5818fb8f5e971d34be4401dbed089 ['66bb83086d8506219ac25610bfd9f24c58f5818fb8f5e971d34be4401dbed089'] A "this document does not change the definition of any http response codes or otherwise proscribe their use." 139 {82} {82: 0} {147: 0} "A server that supports this protocol is called a ""DNS API server"" to differentiate it from a ""DNS server"" (one that uses the regular DNS protocol). Similarly, a client that supports this protocol is called a ""DNS API client""." "this is meant as simpler alternative to . Its much less of a rewrite . consolidates caching rules by moving text adds a new section header "" HTTP Exchange "" for request and response to fall under moves the wireformat definition out of the http sections by moving text . wdyt ? I 'm not terribly happy with pushing the definition of wireformat to so much later because it will cause HTTP people to maybe miss the salient points . This makes the document feel much more like "" HTTPS around DNS "" instead of "" DNS in HTTPS "" , but I can live with it . Given that there are more knobs and fiddly bits in the HTTP part than in the DNS part , it 's reasonable to put those first . if you 're OK with it , say so here and I 'll accept . I 'll then make some very small editorial changes that I am seeing . I 'm OK with it . Would putting the wire format first help ? FWIW I do n't think HTTP folks are going to miss bits just because of the position in the doc . Thanks all . I think this is on balance a useful change . I 'll fix the bitrot and merge ." 5a3bed2a8eb592a1b08dabe58a2a1445d4fe0ab5 5a3bed2a8eb592a1b08dabe58a2a1445d4fe0ab5 draft-ietf-doh-dns-over-https-latest 0f999c14b18da6253b56925e3142fdf439ac883d1f734c1bc2b267198c35a011 ['b66d70acdebb9d8625eb60ce1dcef448418bee2f031f76440b937dbc6db7f80d', '0f999c14b18da6253b56925e3142fdf439ac883d1f734c1bc2b267198c35a011'] A "a server that supports this protocol is called a ""dns api server"" to differentiate it from a ""dns server"" (one that uses the regular dns protocol). similarly, a client that supports this protocol is called a ""dns api client""." 139 {82} {82: 0} {147: 0} "A server that supports this protocol on one or more URIs is called a ""DNS API server"" to differentiate it from a ""DNS server"" (one that uses the regular DNS protocol). Similarly, a client that supports this protocol is called a ""DNS API client""." "this is meant as simpler alternative to . Its much less of a rewrite . consolidates caching rules by moving text adds a new section header "" HTTP Exchange "" for request and response to fall under moves the wireformat definition out of the http sections by moving text . wdyt ? I 'm not terribly happy with pushing the definition of wireformat to so much later because it will cause HTTP people to maybe miss the salient points . This makes the document feel much more like "" HTTPS around DNS "" instead of "" DNS in HTTPS "" , but I can live with it . Given that there are more knobs and fiddly bits in the HTTP part than in the DNS part , it 's reasonable to put those first . if you 're OK with it , say so here and I 'll accept . I 'll then make some very small editorial changes that I am seeing . I 'm OK with it . Would putting the wire format first help ? FWIW I do n't think HTTP folks are going to miss bits just because of the position in the doc . Thanks all . I think this is on balance a useful change . I 'll fix the bitrot and merge ." 5a3bed2a8eb592a1b08dabe58a2a1445d4fe0ab5 5a3bed2a8eb592a1b08dabe58a2a1445d4fe0ab5 draft-ietf-doh-dns-over-https-latest b66d70acdebb9d8625eb60ce1dcef448418bee2f031f76440b937dbc6db7f80d ['b66d70acdebb9d8625eb60ce1dcef448418bee2f031f76440b937dbc6db7f80d', '0f999c14b18da6253b56925e3142fdf439ac883d1f734c1bc2b267198c35a011'] B "a server that supports this protocol on one or more uris is called a ""dns api server"" to differentiate it from a ""dns server"" (one that uses the regular dns protocol). similarly, a client that supports this protocol is called a ""dns api client""." 138 {} {} {} "The messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, parallelism, priority, and header compression to achieve similar performance. Those features were introduced to HTTP in HTTP/2 RFC7540. Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but may result in very poor performance for many uses cases." "Hello , I am a non - native English speaker but I noticed the following sentence : "" Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but may result in very poor performance for many uses cases . "" Should n't be there at the end ? Wiktionary ( ) suggests that plural of "" use case "" is "" use cases "" , not "" uses cases "" ." 5f71a49a36acfc88c4f9f96c74e65daa37dfd500 5f71a49a36acfc88c4f9f96c74e65daa37dfd500 draft-ietf-doh-dns-over-https-latest 2b3bbae76616ea6a912e31de00986adf48076997a842be2942333c761e3c993e ['336527558701898ab519408eca774c4693027174cfbd8237fe4387ffef414b30', '2b3bbae76616ea6a912e31de00986adf48076997a842be2942333c761e3c993e'] https://en.wiktionary.org/wiki/use_case A "the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, parallelism, priority, and header compression to achieve similar performance. those features were introduced to http in http/2 rfc7540. earlier versions of http are capable of conveying the semantic requirements of doh but may result in very poor performance for many uses cases." 138 {} {} {} "The messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, parallelism, priority, and header compression to achieve similar performance. Those features were introduced to HTTP in HTTP/2 RFC7540. Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but may result in very poor performance." "Hello , I am a non - native English speaker but I noticed the following sentence : "" Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but may result in very poor performance for many uses cases . "" Should n't be there at the end ? Wiktionary ( ) suggests that plural of "" use case "" is "" use cases "" , not "" uses cases "" ." 5f71a49a36acfc88c4f9f96c74e65daa37dfd500 5f71a49a36acfc88c4f9f96c74e65daa37dfd500 draft-ietf-doh-dns-over-https-latest 336527558701898ab519408eca774c4693027174cfbd8237fe4387ffef414b30 ['336527558701898ab519408eca774c4693027174cfbd8237fe4387ffef414b30', '2b3bbae76616ea6a912e31de00986adf48076997a842be2942333c761e3c993e'] https://en.wiktionary.org/wiki/use_case B "the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, parallelism, priority, and header compression to achieve similar performance. those features were introduced to http in http/2 rfc7540. earlier versions of http are capable of conveying the semantic requirements of doh but may result in very poor performance." 133 {65, 67, 36, 73, 10, 75, 113} {67: 0, 73: 1, 65: 2, 36: 2, 113: 2, 10: 3, 75: 3} {133: 0, 131: 1, 61: 2, 128: 2, 199: 2, 6: 3, 33: 3} "The URI template defined in this document is processed without any variables for requests using POST, and with the single variable ""dns"" for requests using GET. The value of the dns parameter is the content of the request (as described in dnswire), encoded with base64url RFC4648." "dns request clarified in From Star Brilliant on the DOH WG mailing list : Issue 1 : I found the phrase "" URI template "" is spelled with a lower "" t "" , where all other occurrences of this phrase is spelled with an upper "" T "" . Is this a typo ? Issue 2 : I suggest we change the word "" and "" in Line 2 to "" or "" , for the reason that an HTTP request could not be both GET and POST ." 87e8d0d22648b623d79017f42a2f978d60800f78 87e8d0d22648b623d79017f42a2f978d60800f78 draft-ietf-doh-dns-over-https-latest 8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45 ['8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', 'e13653288d3287fffb40eb2852ac41840c24dce69e015e8d45bf391ec9642da5'] A "the uri template defined in this document is processed without any variables for requests using post, and with the single variable ""dns"" for requests using get. the value of the dns parameter is the content of the request (as described in dnswire), encoded with base64url rfc4648." 133 {65, 67, 36, 73, 10, 75, 113} {67: 0, 73: 1, 65: 2, 36: 2, 113: 2, 10: 3, 75: 3} {133: 0, 131: 1, 61: 2, 128: 2, 199: 2, 6: 3, 33: 3} "The URI Template defined in this document is processed without any variables when the HTTP method is POST. When the HTTP method is GET the single variable ""dns"" is defined as the content of the request (as described in dnswire), encoded with base64url RFC4648." "dns request clarified in From Star Brilliant on the DOH WG mailing list : Issue 1 : I found the phrase "" URI template "" is spelled with a lower "" t "" , where all other occurrences of this phrase is spelled with an upper "" T "" . Is this a typo ? Issue 2 : I suggest we change the word "" and "" in Line 2 to "" or "" , for the reason that an HTTP request could not be both GET and POST ." 87e8d0d22648b623d79017f42a2f978d60800f78 87e8d0d22648b623d79017f42a2f978d60800f78 draft-ietf-doh-dns-over-https-latest e13653288d3287fffb40eb2852ac41840c24dce69e015e8d45bf391ec9642da5 ['8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', 'e13653288d3287fffb40eb2852ac41840c24dce69e015e8d45bf391ec9642da5'] B "the uri template defined in this document is processed without any variables when the http method is post. when the http method is get the single variable ""dns"" is defined as the content of the request (as described in dnswire), encoded with base64url rfc4648." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} "To make a DNS API query a DNS API client encodes a single DNS query into an HTTP request using either the HTTP GET or POST method and the other requirements of this section. The DNS API server defines the URI used by the request. Configuration and discovery of the URI is done out of band from this protocol." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157 ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] A "to make a dns api query a dns api client encodes a single dns query into an http request using either the http get or post method and the other requirements of this section. the dns api server defines the uri used by the request. configuration and discovery of the uri is done out of band from this protocol." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} DNS API servers MUST implement both POST and GET methods. "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596 ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] A dns api servers must implement both post and get methods. 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} "When using the GET method the URI path MUST contain a query parameter name-value pair QUERYPARAMETER with the name of ""ct"" and a value indicating the media-format used for the dns parameter. The value may either be an explicit media type (e.g., ""ct=application/dns- udpwireformat&dns=..."") or it may be empty. An empty value indicates the default application/dns-udpwireformat type (e.g., ""ct&dns=..."")." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1 ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] A "when using the get method the uri path must contain a query parameter name-value pair queryparameter with the name of ""ct"" and a value indicating the media-format used for the dns parameter. the value may either be an explicit media type (e.g., ""ct=application/dns- udpwireformat&dns=..."") or it may be empty. an empty value indicates the default application/dns-udpwireformat type (e.g., ""ct&dns=..."")." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""dns"". The value of the parameter is the content of the request potentially encoded with base64url RFC4648. Specifications that define media types for use with DOH, such as DNS Wire Format dnswire of this document, MUST indicate if the dns parameter uses base64url encoding." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949 ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] A "when using the get method the uri path must contain a query parameter with the name of ""dns"". the value of the parameter is the content of the request potentially encoded with base64url rfc4648. specifications that define media types for use with doh, such as dns wire format dnswire of this document, must indicate if the dns parameter uses base64url encoding." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} "A DNS API client encodes a single DNS query into an HTTP request using either the HTTP GET or POST method and the other requirements of this section. The DNS API server defines the URI used by the request through the use of a URI Template RFC6570. Configuration and discovery of the URI Template is done out of band from this protocol." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] B "a dns api client encodes a single dns query into an http request using either the http get or post method and the other requirements of this section. the dns api server defines the uri used by the request through the use of a uri template rfc6570. configuration and discovery of the uri template is done out of band from this protocol." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} "The URI template defined in this document is processed without any variables for requests using POST, and with the single variable ""dns"" for requests using GET. The value of the dns parameter is the content of the request (as described in dnswire), encoded with base64url RFC4648." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45 ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] B "the uri template defined in this document is processed without any variables for requests using post, and with the single variable ""dns"" for requests using get. the value of the dns parameter is the content of the request (as described in dnswire), encoded with base64url rfc4648." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} "Future specifications for new media types MUST define the variables used for URI Template processing with this protocol." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25 ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] B "future specifications for new media types must define the variables used for uri template processing with this protocol." 131 {65, 67, 36, 73, 10, 75, 113} {73: 0, 65: 1, 67: 1, 36: 1, 113: 1, 10: 2, 75: 2} {131: 0, 61: 1, 133: 1, 128: 1, 199: 1, 6: 2, 33: 2} DNS API servers MUST implement both the POST and GET methods. "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc ['9217065764669ca6154aa4d916c6a2009f1653790eaf621516da2115d72040f1', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '0edb77f298020b19be301220c9d3dcef7c9d3c5aec681a586781680b8b815d25', 'e84a4ab32483cdf057e6db96822271df71c049cac0872b206a4d06c4e33b5949', '8e87684dd09efcd34aaa51a4230b43ef010df4e6a934a367f985770dd1928a45', '63d1ac1fb603d76941786120b7b07e6b081b079e5826555cdc44f9aa817b242a', '55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596', '3811dd183d16a2254bb96cbf774ca5c83869436e62c7ff5ddd8cae67f3939abc'] B dns api servers must implement both the post and get methods. 131 {71, 72, 76, 77, 48, 55} {48: 0, 76: 1, 72: 2, 77: 2, 55: 2, 71: 3} {131: 0, 196: 1, 144: 2, 122: 2, 199: 2, 142: 3} "When using the GET method, the data payload MUST be encoded with base64url RFC4648 and then placed as a name value pair in the query portion of the URI with name ""dns"". Padding characters for base64url MUST NOT be included." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 7237bd131684d8a7cb1fd89edbf70166080b6d95e9bc4e94507ae28bd677e33f ['7237bd131684d8a7cb1fd89edbf70166080b6d95e9bc4e94507ae28bd677e33f', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e'] A "when using the get method, the data payload must be encoded with base64url rfc4648 and then placed as a name value pair in the query portion of the uri with name ""dns"". padding characters for base64url must not be included." 131 {71, 72, 76, 77, 48, 55} {48: 0, 76: 1, 72: 2, 77: 2, 55: 2, 71: 3} {131: 0, 196: 1, 144: 2, 122: 2, 199: 2, 142: 3} "When using the GET method, the data payload MUST be encoded with base64url RFC4648 and then provided as a variable named ""dns"" to the URI Template expansion. Padding characters for base64url MUST NOT be included." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e ['7237bd131684d8a7cb1fd89edbf70166080b6d95e9bc4e94507ae28bd677e33f', '3fdf45f85d53200abae324058b68ab4ced947d95f7d578719a7117315212f90e'] B "when using the get method, the data payload must be encoded with base64url rfc4648 and then provided as a variable named ""dns"" to the uri template expansion. padding characters for base64url must not be included." 131 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {54: 0, 53: 1, 5: 1, 107: 2, 64: 2, 4: 2, 106: 3, 12: 3, 98: 3, 99: 4} {131: 0, 199: 1, 65: 1, 144: 2, 61: 2, 78: 2, 142: 3, 20: 3} "These examples use a DNS API service located at https://dnsserver.example.net/dns-query to resolve the IN A records." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53 ['795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b'] A "these examples use a dns api service located at https://dnsserver.example.net/dns-query to resolve the in a records." 131 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {54: 0, 53: 1, 5: 1, 107: 2, 64: 2, 4: 2, 106: 3, 12: 3, 98: 3, 99: 4} {131: 0, 199: 1, 65: 1, 144: 2, 61: 2, 78: 2, 142: 3, 20: 3} "These examples use a DNS API service with a URI Template of ""https://dnsserver.example.net/dns-query{?dns}"" to resolve IN A records." "? This probably fell through the cracks when was closed . The current design of the GET - based request uses two fields that work in tandem . carries a content type , carries the query . This is suboptimal from a performance perspective when a new request format is developed , which would effectively discourage adoption of new formats ( though most likely it will just prolong the lifetime of the baseline format ) . A query in the form with an implicit content type would be ideal . New formats for requests could do this : . POST requests ca n't benefit from this sort of approach . Not sure what I think about that . My question is : does this also extend to multi - parameter query types in the future , e.g. ? Yes , I am for having query types yes ! My pre - doh thing was accounting for name+type to allow for that granularity , for example MX resolution ( looking outside web - only context ) Other bits are welcome , recursion for example . yes . I am warming up to this . This still seems silly to me for wireformat because if you can decode the answer , you can encode the query . If someone comes up with a JSON response encoding , the query type will be trivial to describe . what other use cases are you thinking of ? The notion would be to create performance incentives for deploying new media types .. you 're right that for it to work you also need to support the old wireformat type so the new type needs to have an advantage other than "" easier "" . Is that not a possibility - will wireformat always be the most expressive ? How could the wire format not be the most expressive ? DNS does n't have an analog of HTTP headers : it 's always just a message body . tl;dr : reopen please . If you subscribe to the notion that wire format is all you ever need , then we 're done already . Just close the door on that possibility and remove . That would be fine for web+JS cases , because all the ugliness of the wire format can be wrapped with an ergonomic API . On the other hand , you might believe there to be value in formats that can express queries in a less - obtuse format . After all , you ca n't slap every problem with a JSL until it goes away . For instance , replacing a ' . ' with a count is pretty onerous in some clients . After considering this some more , I do n't think that you want to manage this by overloading a request with multiple query types . That might sound like a good idea , but it has complications in implementation ( what if the two queries disagree , how do you ensure cachability , etc .... ) . I really want to consider the interaction with discovery , without actually defining a discovery method . Say that you invent a new scheme that includes the combination of multiple QNAMEs and QTYPEs in the following fashion : . Rather than put that query syntax on the same base URL , you would instead configure clients with a discovery document that lists different query modes : Yes , risks straying into territory , but I do n't think that you need to actually specify discovery . Instead , you can instead structure configuration / discovery differently . That would enable that sort of discovery if someone wanted to go there , but it would n't force it on anyone . that this document has precisely one query syntax for GET that syntax with a single variable input ( drop ) and use a URI template to define the placement of that parameter that an alternative query syntax might be supported , but that the syntax would use a different endpoint and URI template and so would require separate configuration That does n't invalidate the current query format , it only changes the configuration . Servers can provide a URI template for their current syntax ( e.g. , would describe the endpoint that is also usable by clients that use configuration from earlier draft versions ) . I 'm a bit torn here . The proposed architecture is fine . My gut also says that its a layer of indirection that wo n't be used in anyway that is more meaningful than what is defined .. and adds it complexity to understanding what is going on which is definitely an anti - goal . your 3 point definition is actually on the side of "" simple enough "" for me , though I 'm not positive we can write it down in a way that does n't create a lot of questions . I 've made a PR with what works for me but I 'm hesitant to go much further . let me know what you think . I have mixed feelings about changing the wireformat so late , but think this is probably the best way to go ." 81d452c78c42b2297ab00c5854a5a382c55fa4b8 81d452c78c42b2297ab00c5854a5a382c55fa4b8 draft-ietf-doh-dns-over-https-latest 3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b ['795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53', '3aa6d26210cbaa32514239922f447d0957b898083fa65b45c954092c3e3e9b4b'] B "these examples use a dns api service with a uri template of ""https://dnsserver.example.net/dns-query{?dns}"" to resolve in a records." 130 {} {} {} "The DNS API client SHOULD include an HTTP ""Accept:"" request header to say what type of content can be understood in response. The client MUST be prepared to process ""application/dns-udpwireformat"" dnswire responses but MAY process any other type it receives." "The fact that an HTTP server can provide a response in the format of its choice was not well understood by DNS people . This motivates the MTI format for servers . On 2018 - 03 - 25 , sent the following to the list : At the meeting in London last week , we asked the room for opinions on two topics : we have a content - type that is Mandatory to Implement ? Several participants spoke in favor of making the default format ( currently called "" application / dns - udpwireformat "" ) mandatory , at least for servers . the GET and POST modes of operation , which , if any , should be mandatory to implement ? Among the participants , there was consensus in favor of making both GET and POST mandatory to implement for servers . If you would like to voice a dissenting opinion on either point , please reply on this list . Otherwise , the chairs will regard these decisions as the group 's consensus . See PR I think most of this can be handled by the MTI PR . But I will just make a tiny clarification indicating that the server is really able to return whatever it needs to ." 3571c554912c24d3b0c29ebd2c0db2c6854ff8b9 3571c554912c24d3b0c29ebd2c0db2c6854ff8b9 draft-ietf-doh-dns-over-https-latest e5d47b6981764adf32ba05d9a655f0a37630d9a279645af6c0c9f30ec60d9700 ['e5d47b6981764adf32ba05d9a655f0a37630d9a279645af6c0c9f30ec60d9700', '9cbe849c13a376d284eaf5cd4685f8a553968107620239c3ffc79175805e0e15'] A "the dns api client should include an http ""accept:"" request header to say what type of content can be understood in response. the client must be prepared to process ""application/dns-udpwireformat"" dnswire responses but may process any other type it receives." 130 {} {} {} "The DNS API client SHOULD include an HTTP ""Accept"" request header to indicate what type of content can be understood in response. Irrespective of the value of the Accept request header, the client MUST be prepared to process ""application/dns-udpwireformat"" dnswire responses but MAY also process any other type it receives." "The fact that an HTTP server can provide a response in the format of its choice was not well understood by DNS people . This motivates the MTI format for servers . On 2018 - 03 - 25 , sent the following to the list : At the meeting in London last week , we asked the room for opinions on two topics : we have a content - type that is Mandatory to Implement ? Several participants spoke in favor of making the default format ( currently called "" application / dns - udpwireformat "" ) mandatory , at least for servers . the GET and POST modes of operation , which , if any , should be mandatory to implement ? Among the participants , there was consensus in favor of making both GET and POST mandatory to implement for servers . If you would like to voice a dissenting opinion on either point , please reply on this list . Otherwise , the chairs will regard these decisions as the group 's consensus . See PR I think most of this can be handled by the MTI PR . But I will just make a tiny clarification indicating that the server is really able to return whatever it needs to ." 3571c554912c24d3b0c29ebd2c0db2c6854ff8b9 3571c554912c24d3b0c29ebd2c0db2c6854ff8b9 draft-ietf-doh-dns-over-https-latest 9cbe849c13a376d284eaf5cd4685f8a553968107620239c3ffc79175805e0e15 ['e5d47b6981764adf32ba05d9a655f0a37630d9a279645af6c0c9f30ec60d9700', '9cbe849c13a376d284eaf5cd4685f8a553968107620239c3ffc79175805e0e15'] B "the dns api client should include an http ""accept"" request header to indicate what type of content can be understood in response. irrespective of the value of the accept request header, the client must be prepared to process ""application/dns-udpwireformat"" dnswire responses but may also process any other type it receives." 129 {17, 2, 3} {2: 0, 3: 1, 17: 2} {129: 0, 119: 1, 40: 2} "Some HTTPS client implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DNS API server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of an OCSP RFC6960 server is one example of how this deadlock can happen. To avoid the possibility of deadlock, DNS API servers SHOULD bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC6961 for TLS version 1.2." "Please do n't include advice about TLS extensions that are n't deployed . RFC 6961 is not widely implemented or used . I 'm sad to say , but RFC 6066 is the best reference for OCSP stapling right now . Also , if you are including this sort of advice , advice against using AIA should go there as well . Some browsers chase AIA URLs in certificates , which will fail for similar reasons . ( You can cite Section 4.2.2.1 of RFC 5280 for that bit ) . Given that this has wandered around a bit , could you create a PR with exactly the changes to the Operational Considerations that you think are needed here ?" 6be5b6e1231486e45adb2a265af7a1ba04c59297 6be5b6e1231486e45adb2a265af7a1ba04c59297 draft-ietf-doh-dns-over-https-latest 025724a2d4c419209079c704cde26cb2bd0c1ed9863ffd64459a6d27d9d36499 ['025724a2d4c419209079c704cde26cb2bd0c1ed9863ffd64459a6d27d9d36499', '03db0de24ec655243401305d97f478331d83809d60f53ad769b9a5cdaa29da5d'] A "some https client implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the dns api server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of an ocsp rfc6960 server is one example of how this deadlock can happen. to avoid the possibility of deadlock, dns api servers should bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc6961 for tls version 1.2." 129 {17, 2, 3} {2: 0, 3: 1, 17: 2} {129: 0, 119: 1, 40: 2} "Some HTTPS client implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DNS API server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of OCSP RFC6960 servers or AIA for CRL fetching (RFC5280 Section 4.2.2.1) are examples of how this deadlock can happen. To mitigate the possibility of deadlock, DNS API servers SHOULD bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC6066 Section 8 for TLS version 1.2." "Please do n't include advice about TLS extensions that are n't deployed . RFC 6961 is not widely implemented or used . I 'm sad to say , but RFC 6066 is the best reference for OCSP stapling right now . Also , if you are including this sort of advice , advice against using AIA should go there as well . Some browsers chase AIA URLs in certificates , which will fail for similar reasons . ( You can cite Section 4.2.2.1 of RFC 5280 for that bit ) . Given that this has wandered around a bit , could you create a PR with exactly the changes to the Operational Considerations that you think are needed here ?" 6be5b6e1231486e45adb2a265af7a1ba04c59297 6be5b6e1231486e45adb2a265af7a1ba04c59297 draft-ietf-doh-dns-over-https-latest 03db0de24ec655243401305d97f478331d83809d60f53ad769b9a5cdaa29da5d ['025724a2d4c419209079c704cde26cb2bd0c1ed9863ffd64459a6d27d9d36499', '03db0de24ec655243401305d97f478331d83809d60f53ad769b9a5cdaa29da5d'] B "some https client implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the dns api server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of ocsp rfc6960 servers or aia for crl fetching (rfc5280 section 4.2.2.1) are examples of how this deadlock can happen. to mitigate the possibility of deadlock, dns api servers should bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc6066 section 8 for tls version 1.2." 128 {65, 67, 36, 73, 10, 75, 113} {36: 0, 73: 1, 65: 2, 67: 2, 113: 2, 10: 3, 75: 3} {128: 0, 131: 1, 61: 2, 133: 2, 199: 2, 6: 3, 33: 3} DNS API servers MUST implement both POST and GET methods. "By my unofficial informal tally , there was a strong agreement on application / dns - udpwireformat being mandatory , and both GET and POST mandatory for servers . I made some very minor edits merge if you 're ok with them The fact that an HTTP server can provide a response in the format of its choice was not well understood by DNS people . This motivates the MTI format for servers . On 2018 - 03 - 25 , sent the following to the list : At the meeting in London last week , we asked the room for opinions on two topics : we have a content - type that is Mandatory to Implement ? Several participants spoke in favor of making the default format ( currently called "" application / dns - udpwireformat "" ) mandatory , at least for servers . the GET and POST modes of operation , which , if any , should be mandatory to implement ? Among the participants , there was consensus in favor of making both GET and POST mandatory to implement for servers . If you would like to voice a dissenting opinion on either point , please reply on this list . Otherwise , the chairs will regard these decisions as the group 's consensus . See PR I think most of this can be handled by the MTI PR . But I will just make a tiny clarification indicating that the server is really able to return whatever it needs to . as discussed on list .. 1 ) Make one mandatory - to - implement 2 ) Make both mandatory - to - implement 3 ) Talk about why there is no mandatory - to - implemen At the face - to - face meeting , I thought the sense of the room was servers MUST implement both GET and POST , clients can implement either or both . This needs to be confirmed on the list . See PR My personal take of consensus is that , having heard no opposition to the position of summary above , to go with that ." 179ce71103c1f1d4a96f2fb4c4a496407a3ee30a 179ce71103c1f1d4a96f2fb4c4a496407a3ee30a draft-ietf-doh-dns-over-https-latest 55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596 ['55955b64d9b518827e48918e170d067ba36a119d85dbfae51a581ba253fe4596'] B dns api servers must implement both post and get methods. 128 {32, 101, 102, 103, 40, 44, 45, 31} {44: 0, 101: 1, 102: 2, 45: 3, 103: 4, 40: 5, 32: 6, 31: 7} {128: 0, 142: 1, 144: 2, 169: 3, 199: 4, 174: 5, 157: 6, 63: 7} "In order to maximize interoperability, DNS API clients and DNS API servers MUST support the ""application/dns-udpwireformat"" media type. Other media types MAY be used as defined by HTTP Content Negotiation (RFC7231 Section 3.4)." "By my unofficial informal tally , there was a strong agreement on application / dns - udpwireformat being mandatory , and both GET and POST mandatory for servers . I made some very minor edits merge if you 're ok with them The fact that an HTTP server can provide a response in the format of its choice was not well understood by DNS people . This motivates the MTI format for servers . On 2018 - 03 - 25 , sent the following to the list : At the meeting in London last week , we asked the room for opinions on two topics : we have a content - type that is Mandatory to Implement ? Several participants spoke in favor of making the default format ( currently called "" application / dns - udpwireformat "" ) mandatory , at least for servers . the GET and POST modes of operation , which , if any , should be mandatory to implement ? Among the participants , there was consensus in favor of making both GET and POST mandatory to implement for servers . If you would like to voice a dissenting opinion on either point , please reply on this list . Otherwise , the chairs will regard these decisions as the group 's consensus . See PR I think most of this can be handled by the MTI PR . But I will just make a tiny clarification indicating that the server is really able to return whatever it needs to . as discussed on list .. 1 ) Make one mandatory - to - implement 2 ) Make both mandatory - to - implement 3 ) Talk about why there is no mandatory - to - implemen At the face - to - face meeting , I thought the sense of the room was servers MUST implement both GET and POST , clients can implement either or both . This needs to be confirmed on the list . See PR My personal take of consensus is that , having heard no opposition to the position of summary above , to go with that ." 179ce71103c1f1d4a96f2fb4c4a496407a3ee30a 179ce71103c1f1d4a96f2fb4c4a496407a3ee30a draft-ietf-doh-dns-over-https-latest 691787dee56f6f17a1e77e896482dc026da67a95929e58d3de1237f7274a2325 ['691787dee56f6f17a1e77e896482dc026da67a95929e58d3de1237f7274a2325'] B "in order to maximize interoperability, dns api clients and dns api servers must support the ""application/dns-udpwireformat"" media type. other media types may be used as defined by http content negotiation (rfc7231 section 3.4)." 127 {} {} {} "This protocol does not define any use for the ""original_transport"" optional parameter of the application/dns-udpwireformat media type." "This is to address the issue of draft - ietf - dnsop - dns - wireformat - http needing to say what the original transport was . Covered on the mailing list March 25 and March 30 . What 's the benefit of "" original _ "" over just plain "" transport "" ?" b2f41d8e3dce70cac5d03ebcbe1c1ee5ef10c1e9 b2f41d8e3dce70cac5d03ebcbe1c1ee5ef10c1e9 draft-ietf-doh-dns-over-https-latest 7dac6fa3b0c17636a7096724302b484f0a86f65f153da408c05b4e463a2a72ce ['7dac6fa3b0c17636a7096724302b484f0a86f65f153da408c05b4e463a2a72ce'] B "this protocol does not define any use for the ""original_transport"" optional parameter of the application/dns-udpwireformat media type." 123 {116, 117, 63} {116: 0, 117: 1, 63: 2} {123: 0, 120: 1, 97: 2} "The data payload is the DNS on-the-wire format defined in RFC1035. The format is for DNS over UDP. Note that this is different than the wire format used in RFC7858. Also note that while RFC1035 says ""Messages carried by UDP are restricted to 512 bytes"", that was later updated by RFC6891, and this protocol does not put any limit on the size of DNS on-the-wire format payloads." "Fixes to PR It does n't feel to me like "" beyond what is already in { { RFC1035 } } "" really clarifies anything , and might just cause additional confusion , as it makes me as a reader wonder , "" oh well what 's in 1035 ? Guess I have to go searching ... "" If you want to put the specific limit from RFC 1035 in , that 's fine . Well no , because I do n't think there should be a limit . While DNS messages implicitly have a maximum size of 64k in DNS , there 's no reason dns - udpwireformat in HTTPS ca n't surpass that . It would be an interoperability problem if a DNS API server handed a DNS API client a message that it could not handle . It is reasonable to assume that DNS API clients expect to parse messages as if they were DNS messages . This was sloppy on my part : I meant the addition to point to RFC 6891 ; now changed . Right but my point is that something like multiple AXFR messages re - assembled into one unit is still a DNS message even if it exceeds 64k . From a wire - format point of view , it parses exactly the same way . Really ? Ugh . Got ta love this protocol . So , you would prefer that the text says "" this protocol allows DNS - on - the - wire format payloads of any size "" ? Yes , I would . Realistically almost everything is still going to be sub-10k , but I do n't see an intrinsic need to limit it , especially with https channels routinely handling far larger responses . I think and I are now in agreement ; should consider accepting ." d1cd36b93252dbe60bb5a1a25adce35d9a48b991 b9caacc79ced199811fa78d42897bcca5283b321 draft-ietf-doh-dns-over-https-latest fa109f744fe622e0746c9c5739aed6e58a0d4f2d00beb65021750292d1944043 ['fa109f744fe622e0746c9c5739aed6e58a0d4f2d00beb65021750292d1944043', '62ec1893ab55da329f0634e806cc5453c8312488db20173cf2f2ea2975ac43ce'] A "the data payload is the dns on-the-wire format defined in rfc1035. the format is for dns over udp. note that this is different than the wire format used in rfc7858. also note that while rfc1035 says ""messages carried by udp are restricted to 512 bytes"", that was later updated by rfc6891, and this protocol does not put any limit on the size of dns on-the-wire format payloads." 123 {116, 117, 63} {116: 0, 117: 1, 63: 2} {123: 0, 120: 1, 97: 2} "The data payload is the DNS on-the-wire format defined in RFC1035. The format is for DNS over UDP. Note that this is different than the wire format used in RFC7858. Also note that while RFC1035 says ""Messages carried by UDP are restricted to 512 bytes"", that was later updated by RFC6891, and this protocol allows DNS on-the-wire format payloads of any size." "Fixes to PR It does n't feel to me like "" beyond what is already in { { RFC1035 } } "" really clarifies anything , and might just cause additional confusion , as it makes me as a reader wonder , "" oh well what 's in 1035 ? Guess I have to go searching ... "" If you want to put the specific limit from RFC 1035 in , that 's fine . Well no , because I do n't think there should be a limit . While DNS messages implicitly have a maximum size of 64k in DNS , there 's no reason dns - udpwireformat in HTTPS ca n't surpass that . It would be an interoperability problem if a DNS API server handed a DNS API client a message that it could not handle . It is reasonable to assume that DNS API clients expect to parse messages as if they were DNS messages . This was sloppy on my part : I meant the addition to point to RFC 6891 ; now changed . Right but my point is that something like multiple AXFR messages re - assembled into one unit is still a DNS message even if it exceeds 64k . From a wire - format point of view , it parses exactly the same way . Really ? Ugh . Got ta love this protocol . So , you would prefer that the text says "" this protocol allows DNS - on - the - wire format payloads of any size "" ? Yes , I would . Realistically almost everything is still going to be sub-10k , but I do n't see an intrinsic need to limit it , especially with https channels routinely handling far larger responses . I think and I are now in agreement ; should consider accepting ." d1cd36b93252dbe60bb5a1a25adce35d9a48b991 b9caacc79ced199811fa78d42897bcca5283b321 draft-ietf-doh-dns-over-https-latest 62ec1893ab55da329f0634e806cc5453c8312488db20173cf2f2ea2975ac43ce ['fa109f744fe622e0746c9c5739aed6e58a0d4f2d00beb65021750292d1944043', '62ec1893ab55da329f0634e806cc5453c8312488db20173cf2f2ea2975ac43ce'] B "the data payload is the dns on-the-wire format defined in rfc1035. the format is for dns over udp. note that this is different than the wire format used in rfc7858. also note that while rfc1035 says ""messages carried by udp are restricted to 512 bytes"", that was later updated by rfc6891, and this protocol allows dns on-the-wire format payloads of any size." 123 {111, 112, 21, 122, 124} {112: 0, 124: 1, 111: 1, 122: 2, 21: 3} {123: 0, 170: 1, 120: 1, 199: 2, 114: 3} "If a DNS API server responds to a DNS API client with a DNS message that has the TC (truncation) bit set in the header, that indicates that the DNS API server was not able to retrieve a full answer for the query over TCP and is providing the best answer it could get. This protocol does not require that a DNS API server that cannot get an untruncated answer send back such an answer; it can instead send back an HTTP error to indicate that it cannot give a useful answer." "Fixes to PR It does n't feel to me like "" beyond what is already in { { RFC1035 } } "" really clarifies anything , and might just cause additional confusion , as it makes me as a reader wonder , "" oh well what 's in 1035 ? Guess I have to go searching ... "" If you want to put the specific limit from RFC 1035 in , that 's fine . Well no , because I do n't think there should be a limit . While DNS messages implicitly have a maximum size of 64k in DNS , there 's no reason dns - udpwireformat in HTTPS ca n't surpass that . It would be an interoperability problem if a DNS API server handed a DNS API client a message that it could not handle . It is reasonable to assume that DNS API clients expect to parse messages as if they were DNS messages . This was sloppy on my part : I meant the addition to point to RFC 6891 ; now changed . Right but my point is that something like multiple AXFR messages re - assembled into one unit is still a DNS message even if it exceeds 64k . From a wire - format point of view , it parses exactly the same way . Really ? Ugh . Got ta love this protocol . So , you would prefer that the text says "" this protocol allows DNS - on - the - wire format payloads of any size "" ? Yes , I would . Realistically almost everything is still going to be sub-10k , but I do n't see an intrinsic need to limit it , especially with https channels routinely handling far larger responses . I think and I are now in agreement ; should consider accepting ." d1cd36b93252dbe60bb5a1a25adce35d9a48b991 b9caacc79ced199811fa78d42897bcca5283b321 draft-ietf-doh-dns-over-https-latest ebb2845cd7c513a5dd05f11c313c28bdc133bbfc1c0ee9b7f32c070eda1faf1a ['ebb2845cd7c513a5dd05f11c313c28bdc133bbfc1c0ee9b7f32c070eda1faf1a', 'ff22f0bfecba010f8dfe93c101d4b4893b7189fad99eb0ee2ffa0052b17c386d'] A "if a dns api server responds to a dns api client with a dns message that has the tc (truncation) bit set in the header, that indicates that the dns api server was not able to retrieve a full answer for the query over tcp and is providing the best answer it could get. this protocol does not require that a dns api server that cannot get an untruncated answer send back such an answer; it can instead send back an http error to indicate that it cannot give a useful answer." 123 {111, 112, 21, 122, 124} {112: 0, 124: 1, 111: 1, 122: 2, 21: 3} {123: 0, 170: 1, 120: 1, 199: 2, 114: 3} "If a DNS API server responds to a DNS API client with a DNS message that has the TC (truncation) bit set in the header, that indicates that the DNS API server was not able to retrieve a full answer for the query and is providing the best answer it could get. This protocol does not require that a DNS API server that cannot get an untruncated answer send back such an answer; it can instead send back an HTTP error to indicate that it cannot give a useful answer." "Fixes to PR It does n't feel to me like "" beyond what is already in { { RFC1035 } } "" really clarifies anything , and might just cause additional confusion , as it makes me as a reader wonder , "" oh well what 's in 1035 ? Guess I have to go searching ... "" If you want to put the specific limit from RFC 1035 in , that 's fine . Well no , because I do n't think there should be a limit . While DNS messages implicitly have a maximum size of 64k in DNS , there 's no reason dns - udpwireformat in HTTPS ca n't surpass that . It would be an interoperability problem if a DNS API server handed a DNS API client a message that it could not handle . It is reasonable to assume that DNS API clients expect to parse messages as if they were DNS messages . This was sloppy on my part : I meant the addition to point to RFC 6891 ; now changed . Right but my point is that something like multiple AXFR messages re - assembled into one unit is still a DNS message even if it exceeds 64k . From a wire - format point of view , it parses exactly the same way . Really ? Ugh . Got ta love this protocol . So , you would prefer that the text says "" this protocol allows DNS - on - the - wire format payloads of any size "" ? Yes , I would . Realistically almost everything is still going to be sub-10k , but I do n't see an intrinsic need to limit it , especially with https channels routinely handling far larger responses . I think and I are now in agreement ; should consider accepting ." d1cd36b93252dbe60bb5a1a25adce35d9a48b991 b9caacc79ced199811fa78d42897bcca5283b321 draft-ietf-doh-dns-over-https-latest ff22f0bfecba010f8dfe93c101d4b4893b7189fad99eb0ee2ffa0052b17c386d ['ebb2845cd7c513a5dd05f11c313c28bdc133bbfc1c0ee9b7f32c070eda1faf1a', 'ff22f0bfecba010f8dfe93c101d4b4893b7189fad99eb0ee2ffa0052b17c386d'] B "if a dns api server responds to a dns api client with a dns message that has the tc (truncation) bit set in the header, that indicates that the dns api server was not able to retrieve a full answer for the query and is providing the best answer it could get. this protocol does not require that a dns api server that cannot get an untruncated answer send back such an answer; it can instead send back an http error to indicate that it cannot give a useful answer." 122 {} {} {} "The queries and responses must be able to be flexible enough to express every normal DNS query." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest 6e4a486d2c90da94920b87a2ddc7d7db112308aec568ad1d21346221f8a7899e ['6e4a486d2c90da94920b87a2ddc7d7db112308aec568ad1d21346221f8a7899e', 'c75e12f782b8e1d95247577435d76563a56e8e9eea099342763b06a8f1778620'] A "the queries and responses must be able to be flexible enough to express every normal dns query." 122 {} {} {} "The queries and responses must be able to be flexible enough to express every DNS query that would normally be sent in DNS over UDP (including queries and responses that use DNS extensions, but not those that require multiple responses)." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest c75e12f782b8e1d95247577435d76563a56e8e9eea099342763b06a8f1778620 ['6e4a486d2c90da94920b87a2ddc7d7db112308aec568ad1d21346221f8a7899e', 'c75e12f782b8e1d95247577435d76563a56e8e9eea099342763b06a8f1778620'] B "the queries and responses must be able to be flexible enough to express every dns query that would normally be sent in dns over udp (including queries and responses that use dns extensions, but not those that require multiple responses)." 122 {57, 108, 29} {29: 0, 108: 1, 57: 2} {122: 0, 106: 1, 50: 2} "The protocol must ensure interoperability by specifying a single format for requests and responses that is mandatory to implement. That format must be able to support future modifications to the DNS protocol including the inclusion of one or more EDNS extensions (including those not yet defined)." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest 41e07035c237cf804d93b17a216a27e92a0ba947676c6b064212b1a615809135 ['41e07035c237cf804d93b17a216a27e92a0ba947676c6b064212b1a615809135', '17fc38c9f4f23336e3f2713a185acc1306d9643f395a85f13ae38657789d0aa7'] A "the protocol must ensure interoperability by specifying a single format for requests and responses that is mandatory to implement. that format must be able to support future modifications to the dns protocol including the inclusion of one or more edns extensions (including those not yet defined)." 122 {57, 108, 29} {29: 0, 108: 1, 57: 2} {122: 0, 106: 1, 50: 2} "The protocol must ensure interoperability by specifying a single format for requests and responses that is mandatory to implement. That format must be able to support future modifications to the DNS protocol including the inclusion of one or more EDNS options (including those not yet defined)." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest 17fc38c9f4f23336e3f2713a185acc1306d9643f395a85f13ae38657789d0aa7 ['41e07035c237cf804d93b17a216a27e92a0ba947676c6b064212b1a615809135', '17fc38c9f4f23336e3f2713a185acc1306d9643f395a85f13ae38657789d0aa7'] B "the protocol must ensure interoperability by specifying a single format for requests and responses that is mandatory to implement. that format must be able to support future modifications to the dns protocol including the inclusion of one or more edns options (including those not yet defined)." 122 {71, 72, 76, 77, 48, 55} {77: 0, 76: 1, 48: 2, 72: 2, 55: 2, 71: 3} {122: 0, 196: 1, 131: 2, 144: 2, 199: 2, 142: 3} "DNS API clients using the DNS wire format MAY have one or more EDNS extensions RFC6891 in the request." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest fa38547c33ce4469c82129ec9acec1371dc832969996204b4bef2a94c31261b3 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', 'fa38547c33ce4469c82129ec9acec1371dc832969996204b4bef2a94c31261b3'] A "dns api clients using the dns wire format may have one or more edns extensions rfc6891 in the request." 122 {71, 72, 76, 77, 48, 55} {77: 0, 76: 1, 48: 2, 72: 2, 55: 2, 71: 3} {122: 0, 196: 1, 131: 2, 144: 2, 199: 2, 142: 3} "DNS API clients using the DNS wire format MAY have one or more EDNS options RFC6891 in the request." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7 ['a4069eab99b796f34100b8e4770e7f39ee79214b64e0c25b8b143d89e7f4fed7', 'fa38547c33ce4469c82129ec9acec1371dc832969996204b4bef2a94c31261b3'] B "dns api clients using the dns wire format may have one or more edns options rfc6891 in the request." 122 {38, 70, 19, 20, 118, 94} {38: 0, 118: 1, 70: 2, 19: 2, 94: 3, 20: 4} {122: 0, 142: 1, 144: 2, 78: 2, 43: 3, 1: 4} "The DNS response for ""application/dns-udpwireformat"" in dnswire MAY have one or more EDNS extensions, depending on the extension definition of the extensions given in the DNS request." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest 58f1420fd001e829fe2039ff28eb388fa4d54fe930d9f335fb4912e86be1256c ['58f1420fd001e829fe2039ff28eb388fa4d54fe930d9f335fb4912e86be1256c', 'fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833'] A "the dns response for ""application/dns-udpwireformat"" in dnswire may have one or more edns extensions, depending on the extension definition of the extensions given in the dns request." 122 {38, 70, 19, 20, 118, 94} {38: 0, 118: 1, 70: 2, 19: 2, 94: 3, 20: 4} {122: 0, 142: 1, 144: 2, 78: 2, 43: 3, 1: 4} "The DNS response for ""application/dns-udpwireformat"" in dnswire MAY have one or more EDNS options, depending on the extension definition of the extensions given in the DNS request." my previous ( deleted now ) comment was in error in case you get a copy of that in email . sorry - git skillz failure on my part . 324823f70e2ee326ed26f18b296c4ffdfca50d9e 324823f70e2ee326ed26f18b296c4ffdfca50d9e draft-ietf-doh-dns-over-https-latest fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833 ['58f1420fd001e829fe2039ff28eb388fa4d54fe930d9f335fb4912e86be1256c', 'fa3960156ddc3b6f3a99a16ca6856cc3c2881b815a473c3edf77c8ca8e3e1833'] B "the dns response for ""application/dns-udpwireformat"" in dnswire may have one or more edns options, depending on the extension definition of the extensions given in the dns request." 120 {116, 117, 63} {117: 0, 116: 1, 63: 1} {120: 0, 123: 1, 97: 1} "The data payload is the DNS on-the-wire format defined in RFC1035. The format is for DNS over UDP. (Note that this is different than the wire format used in RFC7858." "Related to this is a strict improvement . let 's merge and we can adjust further if there is feedback No limit ? Not even the UDP 64 k limit ? I would delete the "" over TCP "" in "" DNS API server was not able to retrieve a full answer for the query over TCP "" The DoH server is supposed to do its best to get the fuil answer , it does not mean we force TCP . Good catches . Will update with I do n't see a compelling technical reason to set a limit . Whether or not RFC 1035 has this limit , DOH should not . The relevant text in RFC 1035 is : or UDP headers ) . Longer messages are truncated and the TC bit is set in the header . If the DOH protocol is referred to as "" per RFC 1035 "" then readers of the spec would be entitled to believe that the 512 byte limit applies unless explicitly documented otherwise . Hello ! as quite a bunch of records are indeed 512 .... some companies set 1024~4096 as operational limit on forwarders and routers . Maybe it could be set in draft to 1024 , to avoid jumboframes at all in the picture ? Maybe not have foreseen the increasing size of messaging we then experienced lately ( along with Add and Auth getting bigger for redundancy ) ? no doubts on it , people are creative about what they can put inside DNS : geo - informations in LOC type , or surf web within TXT records , PGP keys ... I can personally get along most of times with 50 - 100 bytes average packet size in a simple browsing session , but carrying DNSSEC or even just an SPF will make the use of TCP on DNS backend more and more and more exclusive ... I am not saying is any bad / good . We risk limiting the "" native DNS "" initial use of UDP ( which has probably grown after 1035 initial needs ) by reverting back to quite an "" old "" setting . I am not sure everyone today does truncate on exact 512 , I might be very wrong . Little higher limit can still be beneficial to transport , compression , stateless / stateful costs etc . By the way , one of today 's standing argument by Sir , is indeed the present limit on the size of datagrams , just to make things funnier ! Making a parallel with the transport layer , we do not want to drive trough TCP - only world as we had also mentioned QUIC and others . Another parallel on semantics , we almost mention just because comes after HTTP2 but nothing lower than 2 . So why shall we not leave that choice OPEN to implementors , to set the desired truncating threshold starting at 512 an up ? Are there references to present RFC on the topic to help reasoning or delegate the choice on this ? The truncate decision for an ANSWER having size threshold is taken only by server , client will accept and parse so no impact ? Can we try to elaborate on this ? Just brainstorming ... Goodness no , not the least of which being because transport issues are clearly out of scope for this work . The truncate decision for an ANSWER having size threshold is taken only by server , client will accept and parse so no impact ? Can we try to elaborate on this ? The server may have gotten a truncated response from an an authoritative name server , so the client always need to expect an truncated response ( except in some local uses of DNS ) . YES , Seems like something that should be elaborated on . Proposed fix at .. we can reopen this issue if necessary reopen to consider" d86c6a4409bf57f600421248c808b99be4db7df3 d86c6a4409bf57f600421248c808b99be4db7df3 draft-ietf-doh-dns-over-https-latest fe9042f1dcc33ce908506467528fbe6c1eee30ed621b81b63c253c41218fc066 ['fa109f744fe622e0746c9c5739aed6e58a0d4f2d00beb65021750292d1944043', 'fe9042f1dcc33ce908506467528fbe6c1eee30ed621b81b63c253c41218fc066'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/123 A "the data payload is the dns on-the-wire format defined in rfc1035. the format is for dns over udp. (note that this is different than the wire format used in rfc7858." 120 {116, 117, 63} {117: 0, 116: 1, 63: 1} {120: 0, 123: 1, 97: 1} "The data payload is the DNS on-the-wire format defined in RFC1035. The format is for DNS over UDP. Note that this is different than the wire format used in RFC7858. Also note that while RFC1035 says ""Messages carried by UDP are restricted to 512 bytes"", that was later updated by RFC6891, and this protocol does not put any limit on the size of DNS on-the-wire format payloads." "Related to this is a strict improvement . let 's merge and we can adjust further if there is feedback No limit ? Not even the UDP 64 k limit ? I would delete the "" over TCP "" in "" DNS API server was not able to retrieve a full answer for the query over TCP "" The DoH server is supposed to do its best to get the fuil answer , it does not mean we force TCP . Good catches . Will update with I do n't see a compelling technical reason to set a limit . Whether or not RFC 1035 has this limit , DOH should not . The relevant text in RFC 1035 is : or UDP headers ) . Longer messages are truncated and the TC bit is set in the header . If the DOH protocol is referred to as "" per RFC 1035 "" then readers of the spec would be entitled to believe that the 512 byte limit applies unless explicitly documented otherwise . Hello ! as quite a bunch of records are indeed 512 .... some companies set 1024~4096 as operational limit on forwarders and routers . Maybe it could be set in draft to 1024 , to avoid jumboframes at all in the picture ? Maybe not have foreseen the increasing size of messaging we then experienced lately ( along with Add and Auth getting bigger for redundancy ) ? no doubts on it , people are creative about what they can put inside DNS : geo - informations in LOC type , or surf web within TXT records , PGP keys ... I can personally get along most of times with 50 - 100 bytes average packet size in a simple browsing session , but carrying DNSSEC or even just an SPF will make the use of TCP on DNS backend more and more and more exclusive ... I am not saying is any bad / good . We risk limiting the "" native DNS "" initial use of UDP ( which has probably grown after 1035 initial needs ) by reverting back to quite an "" old "" setting . I am not sure everyone today does truncate on exact 512 , I might be very wrong . Little higher limit can still be beneficial to transport , compression , stateless / stateful costs etc . By the way , one of today 's standing argument by Sir , is indeed the present limit on the size of datagrams , just to make things funnier ! Making a parallel with the transport layer , we do not want to drive trough TCP - only world as we had also mentioned QUIC and others . Another parallel on semantics , we almost mention just because comes after HTTP2 but nothing lower than 2 . So why shall we not leave that choice OPEN to implementors , to set the desired truncating threshold starting at 512 an up ? Are there references to present RFC on the topic to help reasoning or delegate the choice on this ? The truncate decision for an ANSWER having size threshold is taken only by server , client will accept and parse so no impact ? Can we try to elaborate on this ? Just brainstorming ... Goodness no , not the least of which being because transport issues are clearly out of scope for this work . The truncate decision for an ANSWER having size threshold is taken only by server , client will accept and parse so no impact ? Can we try to elaborate on this ? The server may have gotten a truncated response from an an authoritative name server , so the client always need to expect an truncated response ( except in some local uses of DNS ) . YES , Seems like something that should be elaborated on . Proposed fix at .. we can reopen this issue if necessary reopen to consider" d86c6a4409bf57f600421248c808b99be4db7df3 d86c6a4409bf57f600421248c808b99be4db7df3 draft-ietf-doh-dns-over-https-latest fa109f744fe622e0746c9c5739aed6e58a0d4f2d00beb65021750292d1944043 ['fa109f744fe622e0746c9c5739aed6e58a0d4f2d00beb65021750292d1944043', 'fe9042f1dcc33ce908506467528fbe6c1eee30ed621b81b63c253c41218fc066'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/123 B "the data payload is the dns on-the-wire format defined in rfc1035. the format is for dns over udp. note that this is different than the wire format used in rfc7858. also note that while rfc1035 says ""messages carried by udp are restricted to 512 bytes"", that was later updated by rfc6891, and this protocol does not put any limit on the size of dns on-the-wire format payloads." 120 {111, 112, 21, 122, 124} {111: 0, 112: 1, 124: 2, 122: 3, 21: 4} {120: 0, 123: 1, 170: 2, 199: 3, 114: 4} "If a DNS API server responds to a DNS API client with a DNS message that has the TC (truncation) bit set in the header, that indicates that the DNS API server was not able to retrieve a full answer for the query over TCP and is providing the best answer it could get. This protocol does not require that a DNS API server that cannot get an untruncated answer send back such an answer; it can instead send back an HTTP error to indicate that it cannot give a useful answer." "Related to this is a strict improvement . let 's merge and we can adjust further if there is feedback No limit ? Not even the UDP 64 k limit ? I would delete the "" over TCP "" in "" DNS API server was not able to retrieve a full answer for the query over TCP "" The DoH server is supposed to do its best to get the fuil answer , it does not mean we force TCP . Good catches . Will update with I do n't see a compelling technical reason to set a limit . Whether or not RFC 1035 has this limit , DOH should not . The relevant text in RFC 1035 is : or UDP headers ) . Longer messages are truncated and the TC bit is set in the header . If the DOH protocol is referred to as "" per RFC 1035 "" then readers of the spec would be entitled to believe that the 512 byte limit applies unless explicitly documented otherwise . Hello ! as quite a bunch of records are indeed 512 .... some companies set 1024~4096 as operational limit on forwarders and routers . Maybe it could be set in draft to 1024 , to avoid jumboframes at all in the picture ? Maybe not have foreseen the increasing size of messaging we then experienced lately ( along with Add and Auth getting bigger for redundancy ) ? no doubts on it , people are creative about what they can put inside DNS : geo - informations in LOC type , or surf web within TXT records , PGP keys ... I can personally get along most of times with 50 - 100 bytes average packet size in a simple browsing session , but carrying DNSSEC or even just an SPF will make the use of TCP on DNS backend more and more and more exclusive ... I am not saying is any bad / good . We risk limiting the "" native DNS "" initial use of UDP ( which has probably grown after 1035 initial needs ) by reverting back to quite an "" old "" setting . I am not sure everyone today does truncate on exact 512 , I might be very wrong . Little higher limit can still be beneficial to transport , compression , stateless / stateful costs etc . By the way , one of today 's standing argument by Sir , is indeed the present limit on the size of datagrams , just to make things funnier ! Making a parallel with the transport layer , we do not want to drive trough TCP - only world as we had also mentioned QUIC and others . Another parallel on semantics , we almost mention just because comes after HTTP2 but nothing lower than 2 . So why shall we not leave that choice OPEN to implementors , to set the desired truncating threshold starting at 512 an up ? Are there references to present RFC on the topic to help reasoning or delegate the choice on this ? The truncate decision for an ANSWER having size threshold is taken only by server , client will accept and parse so no impact ? Can we try to elaborate on this ? Just brainstorming ... Goodness no , not the least of which being because transport issues are clearly out of scope for this work . The truncate decision for an ANSWER having size threshold is taken only by server , client will accept and parse so no impact ? Can we try to elaborate on this ? The server may have gotten a truncated response from an an authoritative name server , so the client always need to expect an truncated response ( except in some local uses of DNS ) . YES , Seems like something that should be elaborated on . Proposed fix at .. we can reopen this issue if necessary reopen to consider" d86c6a4409bf57f600421248c808b99be4db7df3 d86c6a4409bf57f600421248c808b99be4db7df3 draft-ietf-doh-dns-over-https-latest ebb2845cd7c513a5dd05f11c313c28bdc133bbfc1c0ee9b7f32c070eda1faf1a ['ebb2845cd7c513a5dd05f11c313c28bdc133bbfc1c0ee9b7f32c070eda1faf1a'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/123 B "if a dns api server responds to a dns api client with a dns message that has the tc (truncation) bit set in the header, that indicates that the dns api server was not able to retrieve a full answer for the query over tcp and is providing the best answer it could get. this protocol does not require that a dns api server that cannot get an untruncated answer send back such an answer; it can instead send back an http error to indicate that it cannot give a useful answer." 119 {17, 2, 3} {3: 0, 17: 1, 2: 1} {119: 0, 40: 1, 129: 1} "Many HTTPS implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DNS API server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of an OCSP RFC6960 server is one example of how this can happen. DNS API servers SHOULD utilize OCSP Stapling RFC6961 to provide the client with certificate revocation information that does not require contacting a third party." "hmm . we do need to fix this - but the proposed fix removes a normative should and introduces tls 1.3 which creates a publishing dependency . The dep is rather minor , but the benefit to DoH even more minor . let me see if I can come up with a fix that is just a SHOULD inline / staple We can not say SHOULD use 6961 because that is not allowed in TLS 1.3 . I purposely did n't make a publishing dependency to TLS 1.3 by naming it but not giving a link ; every reader of this document will know what we are talking about . I understand the problem . We can probably say something like SHOULD generically staple or inline ( such as 6961 where it applies ) . tls 1.3 does do stapling , it just does n't do it via 6961 . wdyt of that ? Reworded again , and back to you "" DNS API servers SHOULD utilize OCSP Stapling [ RFC6961 ] to provide the client with certificate revocation information that does not require contacting a third party . "" TLS 1.3 will probably be published before DOH and TLS 1.3 deprecates / obsoletes RFC 6961 . TLS 1.3 uses only RFC 6066 . Maybe you should refer to both RFC 6066 and RFC 6961 . Good catch ! Open as PR ." 24426c151959b264b0e57bf144da0918bd03ef8f 2ba345b0f4f5e3288fbc91def83d0fbc2594694e draft-ietf-doh-dns-over-https-latest 1d9ef1733bb8185e7f68e954808847031442cc43eb03659ac1967d02f4b498e4 ['025724a2d4c419209079c704cde26cb2bd0c1ed9863ffd64459a6d27d9d36499', '1d9ef1733bb8185e7f68e954808847031442cc43eb03659ac1967d02f4b498e4'] A "many https implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the dns api server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of an ocsp rfc6960 server is one example of how this can happen. dns api servers should utilize ocsp stapling rfc6961 to provide the client with certificate revocation information that does not require contacting a third party." 119 {17, 2, 3} {3: 0, 17: 1, 2: 1} {119: 0, 40: 1, 129: 1} "Some HTTPS client implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DNS API server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of an OCSP RFC6960 server is one example of how this deadlock can happen. To avoid the possibility of deadlock, DNS API servers SHOULD bundle the certificate status as part of the handshake using a mechanism appropriate to the version of TLS, such as using RFC6961 for TLS version 1.2." "hmm . we do need to fix this - but the proposed fix removes a normative should and introduces tls 1.3 which creates a publishing dependency . The dep is rather minor , but the benefit to DoH even more minor . let me see if I can come up with a fix that is just a SHOULD inline / staple We can not say SHOULD use 6961 because that is not allowed in TLS 1.3 . I purposely did n't make a publishing dependency to TLS 1.3 by naming it but not giving a link ; every reader of this document will know what we are talking about . I understand the problem . We can probably say something like SHOULD generically staple or inline ( such as 6961 where it applies ) . tls 1.3 does do stapling , it just does n't do it via 6961 . wdyt of that ? Reworded again , and back to you "" DNS API servers SHOULD utilize OCSP Stapling [ RFC6961 ] to provide the client with certificate revocation information that does not require contacting a third party . "" TLS 1.3 will probably be published before DOH and TLS 1.3 deprecates / obsoletes RFC 6961 . TLS 1.3 uses only RFC 6066 . Maybe you should refer to both RFC 6066 and RFC 6961 . Good catch ! Open as PR ." 24426c151959b264b0e57bf144da0918bd03ef8f 2ba345b0f4f5e3288fbc91def83d0fbc2594694e draft-ietf-doh-dns-over-https-latest 025724a2d4c419209079c704cde26cb2bd0c1ed9863ffd64459a6d27d9d36499 ['025724a2d4c419209079c704cde26cb2bd0c1ed9863ffd64459a6d27d9d36499', '1d9ef1733bb8185e7f68e954808847031442cc43eb03659ac1967d02f4b498e4'] B "some https client implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the dns api server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of an ocsp rfc6960 server is one example of how this deadlock can happen. to avoid the possibility of deadlock, dns api servers should bundle the certificate status as part of the handshake using a mechanism appropriate to the version of tls, such as using rfc6961 for tls version 1.2." 116 {49, 84, 85} {85: 0, 49: 1, 84: 1} {116: 0, 89: 1, 187: 1} "A client can use DNS over HTTPS as one of multiple mechanisms to obtain DNS data. If a client of this protocol encounters an HTTP error after sending a DNS query, and then falls back to a different DNS retrieval mechanism, doing so can weaken the privacy expected by the user of the client." "authenticity The potential for misunderstanding here is great . If we consider authenticity to be stated as a goal , we would need more explicit about the properties that this provides . Specifically , this does not provide authenticity with respect to responses , because that is the purpose of DNSSEC . Yes , that is true . Thinking about it , I think a web application developer ( one of the targets ) reading this draft might not fully understand the differences between authenticity at the HTTP layer and authenticity at the DNS layer . One of the stated features of is authenticated channels that guarantee data integrity . If possible , I think a few more sentences elaborating on this topic would not hurt . I 'm going to close this because we have no idea what kind of authentication that a user would expect , whereas we know pretty well what kind of privacy they expect . I think there might be room for something different along these lines . DNS hijack from your configured DNS server is currently a major problem , and DoH does allow you to authenticate whether or not you are talking to your configured recursive ( while not saying anything about e2e ) . I 'm going to reopen to track .." 913391d59528bd82b0b1aefeac095a822d177582 913391d59528bd82b0b1aefeac095a822d177582 draft-ietf-doh-dns-over-https-latest 72d1aa8c65389efe1cf83588d7d4893742505d4335b4565fac9e7256374d7d8e ['be312f4e116d747b2945a7685e0a3846619c91137fc5169e4f16adc33e4b568e', '72d1aa8c65389efe1cf83588d7d4893742505d4335b4565fac9e7256374d7d8e'] A "a client can use dns over https as one of multiple mechanisms to obtain dns data. if a client of this protocol encounters an http error after sending a dns query, and then falls back to a different dns retrieval mechanism, doing so can weaken the privacy expected by the user of the client." 116 {49, 84, 85} {85: 0, 49: 1, 84: 1} {116: 0, 89: 1, 187: 1} "A client can use DNS over HTTPS as one of multiple mechanisms to obtain DNS data. If a client of this protocol encounters an HTTP error after sending a DNS query, and then falls back to a different DNS retrieval mechanism, doing so can weaken the privacy and authenticity expected by the user of the client." "authenticity The potential for misunderstanding here is great . If we consider authenticity to be stated as a goal , we would need more explicit about the properties that this provides . Specifically , this does not provide authenticity with respect to responses , because that is the purpose of DNSSEC . Yes , that is true . Thinking about it , I think a web application developer ( one of the targets ) reading this draft might not fully understand the differences between authenticity at the HTTP layer and authenticity at the DNS layer . One of the stated features of is authenticated channels that guarantee data integrity . If possible , I think a few more sentences elaborating on this topic would not hurt . I 'm going to close this because we have no idea what kind of authentication that a user would expect , whereas we know pretty well what kind of privacy they expect . I think there might be room for something different along these lines . DNS hijack from your configured DNS server is currently a major problem , and DoH does allow you to authenticate whether or not you are talking to your configured recursive ( while not saying anything about e2e ) . I 'm going to reopen to track .." 913391d59528bd82b0b1aefeac095a822d177582 913391d59528bd82b0b1aefeac095a822d177582 draft-ietf-doh-dns-over-https-latest be312f4e116d747b2945a7685e0a3846619c91137fc5169e4f16adc33e4b568e ['be312f4e116d747b2945a7685e0a3846619c91137fc5169e4f16adc33e4b568e', '72d1aa8c65389efe1cf83588d7d4893742505d4335b4565fac9e7256374d7d8e'] B "a client can use dns over https as one of multiple mechanisms to obtain dns data. if a client of this protocol encounters an http error after sending a dns query, and then falls back to a different dns retrieval mechanism, doing so can weaken the privacy and authenticity expected by the user of the client." 114 {90, 60} {90: 0, 60: 1} {114: 0, 101: 1} "A DOH API client may utilize a hierarchy of caches that include both HTTP and DNS specific caches. HTTP cache entries may be bypassed with HTTP mechanisms such as the ""Cache-Control no-cache"" directive; however DNS caches do not have a similar mechanism." "Yipes , I thought I had found these in the last round . Thanks ! Even though the Terminology section contains "" DNS API client "" and "" DNS API server "" respectively , the version of the draft still contains a few instances of "" DOH client "" and "" DOH server "" . Not sure that 's intended . As well as one instance of "" DOH API client "" in Section 6.1" 8e3e59e07454a3572105be4ffc802670877f8653 8e3e59e07454a3572105be4ffc802670877f8653 draft-ietf-doh-dns-over-https-latest c4b5bd79f7946ee20ea09a1fa8e24f7f9dd468b162ecde1f046514d00d826e05 ['c69d2acfa8a7a04236f2519c769343d29c240eb72d371e8d1c2b486e19f52544', 'c4b5bd79f7946ee20ea09a1fa8e24f7f9dd468b162ecde1f046514d00d826e05'] A "a doh api client may utilize a hierarchy of caches that include both http and dns specific caches. http cache entries may be bypassed with http mechanisms such as the ""cache-control no-cache"" directive; however dns caches do not have a similar mechanism." 114 {90, 60} {90: 0, 60: 1} {114: 0, 101: 1} "A DNS API client may utilize a hierarchy of caches that include both HTTP and DNS specific caches. HTTP cache entries may be bypassed with HTTP mechanisms such as the ""Cache-Control no-cache"" directive; however DNS caches do not have a similar mechanism." "Yipes , I thought I had found these in the last round . Thanks ! Even though the Terminology section contains "" DNS API client "" and "" DNS API server "" respectively , the version of the draft still contains a few instances of "" DOH client "" and "" DOH server "" . Not sure that 's intended . As well as one instance of "" DOH API client "" in Section 6.1" 8e3e59e07454a3572105be4ffc802670877f8653 8e3e59e07454a3572105be4ffc802670877f8653 draft-ietf-doh-dns-over-https-latest c69d2acfa8a7a04236f2519c769343d29c240eb72d371e8d1c2b486e19f52544 ['c69d2acfa8a7a04236f2519c769343d29c240eb72d371e8d1c2b486e19f52544', 'c4b5bd79f7946ee20ea09a1fa8e24f7f9dd468b162ecde1f046514d00d826e05'] B "a dns api client may utilize a hierarchy of caches that include both http and dns specific caches. http cache entries may be bypassed with http mechanisms such as the ""cache-control no-cache"" directive; however dns caches do not have a similar mechanism." 114 {111, 112, 21, 122, 124} {21: 0, 122: 1, 124: 2, 112: 3, 111: 4} {114: 0, 199: 1, 170: 2, 123: 3, 120: 4} "A DNS API client may face a similar bootstrapping problem when the HTTP request needs to resolve the hostname portion of the DNS URI. Just as the address of a traditional DNS nameserver cannot be originally determined from that same server, a DOH client cannot use its DOH server to initially resolve the server's host name into an address. Alternative strategies a client might employ include making the initial resolution part of the configuration, IP based URIs and corresponding IP based certificates for HTTPS, or resolving the DNS API server's hostname via traditional DNS or another DOH server while still authenticating the resulting connection via HTTPS." "Yipes , I thought I had found these in the last round . Thanks ! Even though the Terminology section contains "" DNS API client "" and "" DNS API server "" respectively , the version of the draft still contains a few instances of "" DOH client "" and "" DOH server "" . Not sure that 's intended . As well as one instance of "" DOH API client "" in Section 6.1" 8e3e59e07454a3572105be4ffc802670877f8653 8e3e59e07454a3572105be4ffc802670877f8653 draft-ietf-doh-dns-over-https-latest 845c418c0ea01ff1745a3fea65842fa56c4282a1b0f109485b90277cf96689eb ['2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '845c418c0ea01ff1745a3fea65842fa56c4282a1b0f109485b90277cf96689eb'] A "a dns api client may face a similar bootstrapping problem when the http request needs to resolve the hostname portion of the dns uri. just as the address of a traditional dns nameserver cannot be originally determined from that same server, a doh client cannot use its doh server to initially resolve the server's host name into an address. alternative strategies a client might employ include making the initial resolution part of the configuration, ip based uris and corresponding ip based certificates for https, or resolving the dns api server's hostname via traditional dns or another doh server while still authenticating the resulting connection via https." 114 {111, 112, 21, 122, 124} {21: 0, 122: 1, 124: 2, 112: 3, 111: 4} {114: 0, 199: 1, 170: 2, 123: 3, 120: 4} "A DNS API client may face a similar bootstrapping problem when the HTTP request needs to resolve the hostname portion of the DNS URI. Just as the address of a traditional DNS nameserver cannot be originally determined from that same server, a DNS API client cannot use its DNS API server to initially resolve the server's host name into an address. Alternative strategies a client might employ include making the initial resolution part of the configuration, IP based URIs and corresponding IP based certificates for HTTPS, or resolving the DNS API server's hostname via traditional DNS or another DNS API server while still authenticating the resulting connection via HTTPS." "Yipes , I thought I had found these in the last round . Thanks ! Even though the Terminology section contains "" DNS API client "" and "" DNS API server "" respectively , the version of the draft still contains a few instances of "" DOH client "" and "" DOH server "" . Not sure that 's intended . As well as one instance of "" DOH API client "" in Section 6.1" 8e3e59e07454a3572105be4ffc802670877f8653 8e3e59e07454a3572105be4ffc802670877f8653 draft-ietf-doh-dns-over-https-latest 2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9 ['2e19a290b8a09f371cd564d60d127292736e354a3341173ee105dfe81a5d57f9', '845c418c0ea01ff1745a3fea65842fa56c4282a1b0f109485b90277cf96689eb'] B "a dns api client may face a similar bootstrapping problem when the http request needs to resolve the hostname portion of the dns uri. just as the address of a traditional dns nameserver cannot be originally determined from that same server, a dns api client cannot use its dns api server to initially resolve the server's host name into an address. alternative strategies a client might employ include making the initial resolution part of the configuration, ip based uris and corresponding ip based certificates for https, or resolving the dns api server's hostname via traditional dns or another dns api server while still authenticating the resulting connection via https." 106 {57, 108, 29} {108: 0, 57: 1, 29: 1} {106: 0, 50: 1, 122: 1} "The protocol must ensure interoperable media formats through a mandatory to implement format wherein a query must be able to contain future modifications to the DNS protocol including the inclusion of one or more EDNS extensions (including those not yet defined)." This text does not read very well in its current form . Since it is two ideas , two sentences is best . 4ccac7a72cd0e4b0ddaf45e7e377fee077d114af 4ccac7a72cd0e4b0ddaf45e7e377fee077d114af draft-ietf-doh-dns-over-https-latest dcf53581e74ba5e34951581818a0244659b089eef434ba87c5f33af317ac9cdb ['dcf53581e74ba5e34951581818a0244659b089eef434ba87c5f33af317ac9cdb', '41e07035c237cf804d93b17a216a27e92a0ba947676c6b064212b1a615809135'] A "the protocol must ensure interoperable media formats through a mandatory to implement format wherein a query must be able to contain future modifications to the dns protocol including the inclusion of one or more edns extensions (including those not yet defined)." 106 {57, 108, 29} {108: 0, 57: 1, 29: 1} {106: 0, 50: 1, 122: 1} "The protocol must ensure interoperability by specifying a single format for requests and responses that is mandatory to implement. That format must be able to support future modifications to the DNS protocol including the inclusion of one or more EDNS extensions (including those not yet defined)." This text does not read very well in its current form . Since it is two ideas , two sentences is best . 4ccac7a72cd0e4b0ddaf45e7e377fee077d114af 4ccac7a72cd0e4b0ddaf45e7e377fee077d114af draft-ietf-doh-dns-over-https-latest 41e07035c237cf804d93b17a216a27e92a0ba947676c6b064212b1a615809135 ['dcf53581e74ba5e34951581818a0244659b089eef434ba87c5f33af317ac9cdb', '41e07035c237cf804d93b17a216a27e92a0ba947676c6b064212b1a615809135'] B "the protocol must ensure interoperability by specifying a single format for requests and responses that is mandatory to implement. that format must be able to support future modifications to the dns protocol including the inclusion of one or more edns extensions (including those not yet defined)." 105 {} {} {} "The protocol must allow implementations to use HTTP's content negotiation mechanism." " is a means , not an end . this is a good change . I 'm good if is good ." a5c0c4c243ebb2f934ba374c63545acd56717eef a5c0c4c243ebb2f934ba374c63545acd56717eef draft-ietf-doh-dns-over-https-latest 8514a51ac09b9399df5d0ff9c5c165515b02532b141cb5784e4eed089152822e ['8514a51ac09b9399df5d0ff9c5c165515b02532b141cb5784e4eed089152822e', '5d441a2fb8d46a289d6a51392da45b063d7593856c177983e8eec39270287719'] A "the protocol must allow implementations to use http's content negotiation mechanism." 105 {} {} {} "The protocol must permit the addition of new formats for DNS queries and responses." " is a means , not an end . this is a good change . I 'm good if is good ." a5c0c4c243ebb2f934ba374c63545acd56717eef a5c0c4c243ebb2f934ba374c63545acd56717eef draft-ietf-doh-dns-over-https-latest 5d441a2fb8d46a289d6a51392da45b063d7593856c177983e8eec39270287719 ['8514a51ac09b9399df5d0ff9c5c165515b02532b141cb5784e4eed089152822e', '5d441a2fb8d46a289d6a51392da45b063d7593856c177983e8eec39270287719'] B "the protocol must permit the addition of new formats for dns queries and responses." 104 {105, 110} {105: 0, 110: 1} {104: 0, 99: 1} "The Internet does not always provide end to end reachability for native DNS. On-path network devices may spoof DNS responses, block DNS requests, or just redirect DNS queries to different DNS servers that give less-than-honest answers. These are also sometimes delivered with poor performance or reduced feature sets." "This text is great justification to include in a charter , but this is a protocol specification and it really does n't add much . It seemed like the WG wanted the document to be shorter and more specific , so this feels good ." 21b83dbfea9d5559795c00f0cf99f28a5a09af89 21b83dbfea9d5559795c00f0cf99f28a5a09af89 draft-ietf-doh-dns-over-https-latest 02fc2dc006a5e8528c3233c295ac4d2a2122d6052bf5e005ab5d6bfe3c3b1ff4 ['d641b2492a6422a8696559ab3b8e1848a31af9e99abd23d0012b546af464b1d0', '02fc2dc006a5e8528c3233c295ac4d2a2122d6052bf5e005ab5d6bfe3c3b1ff4'] A "the internet does not always provide end to end reachability for native dns. on-path network devices may spoof dns responses, block dns requests, or just redirect dns queries to different dns servers that give less-than-honest answers. these are also sometimes delivered with poor performance or reduced feature sets." 104 {105, 110} {105: 0, 110: 1} {104: 0, 99: 1} "Over time, there have been many proposals for using HTTP and HTTPS as a substrate for DNS queries and responses. To date, none of those proposals have made it beyond early discussion, partially due to disagreement about what the appropriate formatting should be and partially because they did not follow HTTP best practices." "This text is great justification to include in a charter , but this is a protocol specification and it really does n't add much . It seemed like the WG wanted the document to be shorter and more specific , so this feels good ." 21b83dbfea9d5559795c00f0cf99f28a5a09af89 21b83dbfea9d5559795c00f0cf99f28a5a09af89 draft-ietf-doh-dns-over-https-latest d641b2492a6422a8696559ab3b8e1848a31af9e99abd23d0012b546af464b1d0 ['d641b2492a6422a8696559ab3b8e1848a31af9e99abd23d0012b546af464b1d0', '02fc2dc006a5e8528c3233c295ac4d2a2122d6052bf5e005ab5d6bfe3c3b1ff4'] A "over time, there have been many proposals for using http and https as a substrate for dns queries and responses. to date, none of those proposals have made it beyond early discussion, partially due to disagreement about what the appropriate formatting should be and partially because they did not follow http best practices." 101 {33, 34, 18, 120, 26, 30} {30: 0, 18: 1, 26: 2, 33: 3, 34: 4, 120: 5} {101: 0, 56: 1, 39: 2, 33: 3, 21: 4, 42: 5} "The Answer section of a DNS response contains one or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset is supposed to have the Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs, though it is only possible for the HTTP response to have a single freshness lifetime. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the Resource Record bearing the smallest TTL in the Answer section of the response. Specifically, the HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "feel free to merge if my language works for you I just now see you gave me the go ahead to merge . I 'll fix the conflict and do so ." 18de12f307fa0d90c17dddd7d64f3836ce036a4a 18de12f307fa0d90c17dddd7d64f3836ce036a4a draft-ietf-doh-dns-over-https-latest 8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0 ['48155b6d74f4dcc0d5df2016c050097abb559f0c022648e9aa55a5def1dde304', 'f9a714efcb644357495b1734f998a11fc3fabfa1b657e2d3aff97b82c32db091', '8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0'] A "the answer section of a dns response contains one or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset is supposed to have the time to live (ttl) freshness information. different rrsets in the answer section can have different ttls, though it is only possible for the http response to have a single freshness lifetime. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the resource record bearing the smallest ttl in the answer section of the response. specifically, the http freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." 101 {33, 34, 18, 120, 26, 30} {30: 0, 18: 1, 26: 2, 33: 3, 34: 4, 120: 5} {101: 0, 56: 1, 39: 2, 33: 3, 21: 4, 42: 5} "The Answer section of a DNS response can contain zero or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset has Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs, although it is only possible for the HTTP response to have a single freshness lifetime. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the RRset with the smallest TTL in the Answer section of the response. Specifically, the HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS resource records in the Answer section reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS resoruce record with the smallest TTL in the response." "feel free to merge if my language works for you I just now see you gave me the go ahead to merge . I 'll fix the conflict and do so ." 18de12f307fa0d90c17dddd7d64f3836ce036a4a 18de12f307fa0d90c17dddd7d64f3836ce036a4a draft-ietf-doh-dns-over-https-latest f9a714efcb644357495b1734f998a11fc3fabfa1b657e2d3aff97b82c32db091 ['48155b6d74f4dcc0d5df2016c050097abb559f0c022648e9aa55a5def1dde304', 'f9a714efcb644357495b1734f998a11fc3fabfa1b657e2d3aff97b82c32db091', '8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0'] B "the answer section of a dns response can contain zero or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset has time to live (ttl) freshness information. different rrsets in the answer section can have different ttls, although it is only possible for the http response to have a single freshness lifetime. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the rrset with the smallest ttl in the answer section of the response. specifically, the http freshness lifetime should be set to expire at the same time any of the dns resource records in the answer section reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns resoruce record with the smallest ttl in the response." 101 {33, 34, 18, 120, 26, 30} {30: 0, 18: 1, 26: 2, 33: 3, 34: 4, 120: 5} {101: 0, 56: 1, 39: 2, 33: 3, 21: 4, 42: 5} "If the DNS response has no records in the Answer section, and the DNS response has an SOA record in the Authority section, the response freshness lifetime MUST NOT be greater than the MINIMUM field from that SOA record. Otherwise, the HTTP response MUST set a freshness lifetime (RFC7234 Section 4.2) of 0 by using a mechanism such as ""Cache-Control: no-cache"" (RFC7234 Section 5.2.1.4)." "feel free to merge if my language works for you I just now see you gave me the go ahead to merge . I 'll fix the conflict and do so ." 18de12f307fa0d90c17dddd7d64f3836ce036a4a 18de12f307fa0d90c17dddd7d64f3836ce036a4a draft-ietf-doh-dns-over-https-latest 48155b6d74f4dcc0d5df2016c050097abb559f0c022648e9aa55a5def1dde304 ['48155b6d74f4dcc0d5df2016c050097abb559f0c022648e9aa55a5def1dde304', 'f9a714efcb644357495b1734f998a11fc3fabfa1b657e2d3aff97b82c32db091', '8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0'] B "if the dns response has no records in the answer section, and the dns response has an soa record in the authority section, the response freshness lifetime must not be greater than the minimum field from that soa record. otherwise, the http response must set a freshness lifetime (rfc7234 section 4.2) of 0 by using a mechanism such as ""cache-control: no-cache"" (rfc7234 section 5.2.1.4)." 101 {90, 60} {60: 0, 90: 1} {101: 0, 114: 1} "A DOH API client may utilize a hierarchy of caches that include both HTTP and DNS specific caches. HTTP cache entries may be bypassed with HTTP mechanisms such as the Cache-Control no-cache directive however DNS caches do not have a similar mechanism." "feel free to merge if my language works for you I just now see you gave me the go ahead to merge . I 'll fix the conflict and do so ." 18de12f307fa0d90c17dddd7d64f3836ce036a4a 18de12f307fa0d90c17dddd7d64f3836ce036a4a draft-ietf-doh-dns-over-https-latest 851a9bbb8077ef4c5551dbbd68ed3b09116aa9c82af693a7a59d8a30299a2817 ['851a9bbb8077ef4c5551dbbd68ed3b09116aa9c82af693a7a59d8a30299a2817', 'c4b5bd79f7946ee20ea09a1fa8e24f7f9dd468b162ecde1f046514d00d826e05'] A "a doh api client may utilize a hierarchy of caches that include both http and dns specific caches. http cache entries may be bypassed with http mechanisms such as the cache-control no-cache directive however dns caches do not have a similar mechanism." 101 {90, 60} {60: 0, 90: 1} {101: 0, 114: 1} "A DOH API client may utilize a hierarchy of caches that include both HTTP and DNS specific caches. HTTP cache entries may be bypassed with HTTP mechanisms such as the ""Cache-Control no-cache"" directive; however DNS caches do not have a similar mechanism." "feel free to merge if my language works for you I just now see you gave me the go ahead to merge . I 'll fix the conflict and do so ." 18de12f307fa0d90c17dddd7d64f3836ce036a4a 18de12f307fa0d90c17dddd7d64f3836ce036a4a draft-ietf-doh-dns-over-https-latest c4b5bd79f7946ee20ea09a1fa8e24f7f9dd468b162ecde1f046514d00d826e05 ['851a9bbb8077ef4c5551dbbd68ed3b09116aa9c82af693a7a59d8a30299a2817', 'c4b5bd79f7946ee20ea09a1fa8e24f7f9dd468b162ecde1f046514d00d826e05'] B "a doh api client may utilize a hierarchy of caches that include both http and dns specific caches. http cache entries may be bypassed with http mechanisms such as the ""cache-control no-cache"" directive; however dns caches do not have a similar mechanism." 99 {105, 110} {110: 0, 105: 1} {99: 0, 104: 1} "The Internet does not always provide end to end reachability for native DNS. On-path network devices may spoof DNS responses, block DNS requests, or just redirect DNS queries to different DNS servers that give less-than-honest answers." "on list points out that we do n't mention some pretty key architectural benefits of doh such as authentication , amplification resistance , etc . on list as "" changes for draft - ietf - doh - dns - over - https-03 "" . is meant to address this Hi , following few impressions on advantageous aspects of doh : For sure , the experience aquired in mitigation against hammering attacks on "" HTTP "" layer is much more effective and well - ruled than DDoS mitigation(s ) when comparing to UDP DNS , due to persisting amplification attacks . Many different APIs to expire contents are available on a system layer ( Cloudflare , squid , find / rm ) , many HTTP request headers too on a protocl layer , so shall be widely possible to "" control "" cache validity with non - DNS methods , when system is responsive ( i.e. can take little longer).Compression , modernised with brotli and binary proto . I also sustain the need of a mention to the HEAD method as cache priming or verification method .. Is it desired , interesting , considered any useful by the working group ? ( Stephane brought this up to as a bug , but me too I was considering it since ever)Leveraging CDN caches ( when atomic , i.e. , CF - ID ) is another immediate advantage , but as discussed in the CNAME / A thread there might arise issues with ADDITIONAL / AUTHORITY sections ' validity ( my implementation used to squeeze the two in one consistent response , which is doable in DNS terms , for the sake of the service ... or give shortened responses and iterate trough natural granularity of DNS messaging).Preemptive / ALL caching is less interesting to me personally , but I understand the relatively low effort required by infrastructure to take this extra burden of caching+refreshing it ( we now always desire to get the most records at once ... ) . Might state that too , that DNSSEC interactions will require more storage , or more bandwidth , or refer to it . Talking about Auth / Auth , OAuth or all the SSO you can name , are clearly BIG advantages ( i.e. refuse some clients or geo - locate / displace others ) . I was thinking MAC / DAC reasonings were to fall outside of DNS - over - HTTP draft , IMHO ... is not a "" native "" feature of DNS , as - is ... what is your opinion ? to me is HTTP.OTOH , a more precise Accounting is a native aspect coming straight out of doh ( making accounting easier if ever needed , my old implementation accounted for billing , not a new thing ) . To me , just because some companies do bill on imprecise Netflow stats , accurate tcpdumps on mirror ports , or even weirder things , does n't mean DOH can not state ( mention ) about "" billabilty "" at the net of SYN , SYN / ACK , ACK , retransmissions and RESETS :) :) Having a line like : "" as any other HTTP dialogue , DNS - over - HTTP might offer possibility to have its usage logged and billed "" . For some companies it might be a plus , let 's not tell them they could have done it with bind or plugins :) Traceability ? Is a mention to logging needed ? ( or needs to be implied as any HTTP does?).The detail of the query itself might not be available to authorities via such simplified logs anyway , and that has been made clear during the tables anyway , by regarding POSTs for example . In my previous exp in law enforcment for telco -started with landlines then switched to web- European ISP and hosting providers were obliged / required to provide HTTP and mail logs by investigating authorities ( at their own cost of maintenance , law enforcment might be a dead budget for smaller companies ) . It was 2006 , 2009 then 2016 things have n't much improved , just the sensibility towards the topic but not the smoothness of process and guarantee of security in personal / sensible data treatment , in medium companies not just . Many product offer since day 1 full auditing features , FGA in Oracle for example , simple or complex as be , is just matter of using them :( But why DNS traces are not always taken in consideration by all authorities , or everything becomes "" best - effort "" ? The queries / responses might not be traced / traceable , surely not in post - mortem , only some providers can / do . Not sponsoring any point of view on justice or privacy . ... this for doh is an evolution and an involution : the space needed to account for the long - term - storage might be an argument not to spouse doh . Obviously , filtering out is easy ( ct= ) By mentioning such aspect(s ) we can help answering frequent questions before they 're even been asked : +1 : As we are realising , mixing the experts from 2 different worlds of DNS and HTTP is not easy , so the more dialectical the best ? With clear examples for sure . I just think that people need to think what is DOH and what does and what does n't automagically do ( extra certificates , extra logs , "" slower "" but maybe faster , etc ) so to be able to compare pros / cons comparing to non HTTP / DNS servers / clients . To me is doh is just a transport thing , and I really like it since many years :) sorry for super long post , will copy this to mailing list if you think is useful . I agree with almost everything you say , but I do n't think the protocol doc is the place to got at all of these things at this length . I 'm just trying to give a flavor of the implications of using the http ecosystem - which you go into in much deeper detail . The document has to focus on how you do that . my suggestion would be you raise the concept of a separate applicability / use - case document if you think this is valuable .. you could raise this as part of the charter discussion for DoH. Or you could just work on a really great blog post :) Thank you I will follow all of your suggestions , just a matter of having some time to say something "" useful "" and hopefully , possibly correct . I am still willing to follow all of the discussions around the doh topic , I have a lot to learn from all of you , really proud of your consideration . I understand the logic of defining precise specifications and how - to - implement , proper of drafting RFC . I fully support the criteria of "" referring to "" other standing and specific RFCs by keeping the transversal implications outside of the doc , as it might over - confuse implementors and this is not the goal . Sorry for that , I just wanted to mention some of the possible "" arguments "" or side - effects ( my first time in participating in such exciting discussions and group ) . My congratulations to all of the working group and supporters , for all the efforts being put into this important draft" eb8e329701a1e22b3ee1e7e02f5e229649355f0f eb8e329701a1e22b3ee1e7e02f5e229649355f0f draft-ietf-doh-dns-over-https-latest eaf86e75773cda3db897646dbb2845644406e315215006c451c16225d19eefb9 ['eaf86e75773cda3db897646dbb2845644406e315215006c451c16225d19eefb9', '02fc2dc006a5e8528c3233c295ac4d2a2122d6052bf5e005ab5d6bfe3c3b1ff4'] A "the internet does not always provide end to end reachability for native dns. on-path network devices may spoof dns responses, block dns requests, or just redirect dns queries to different dns servers that give less-than-honest answers." 99 {105, 110} {110: 0, 105: 1} {99: 0, 104: 1} "The Internet does not always provide end to end reachability for native DNS. On-path network devices may spoof DNS responses, block DNS requests, or just redirect DNS queries to different DNS servers that give less-than-honest answers. These are also sometimes delivered with poor performance or reduced feature sets." "on list points out that we do n't mention some pretty key architectural benefits of doh such as authentication , amplification resistance , etc . on list as "" changes for draft - ietf - doh - dns - over - https-03 "" . is meant to address this Hi , following few impressions on advantageous aspects of doh : For sure , the experience aquired in mitigation against hammering attacks on "" HTTP "" layer is much more effective and well - ruled than DDoS mitigation(s ) when comparing to UDP DNS , due to persisting amplification attacks . Many different APIs to expire contents are available on a system layer ( Cloudflare , squid , find / rm ) , many HTTP request headers too on a protocl layer , so shall be widely possible to "" control "" cache validity with non - DNS methods , when system is responsive ( i.e. can take little longer).Compression , modernised with brotli and binary proto . I also sustain the need of a mention to the HEAD method as cache priming or verification method .. Is it desired , interesting , considered any useful by the working group ? ( Stephane brought this up to as a bug , but me too I was considering it since ever)Leveraging CDN caches ( when atomic , i.e. , CF - ID ) is another immediate advantage , but as discussed in the CNAME / A thread there might arise issues with ADDITIONAL / AUTHORITY sections ' validity ( my implementation used to squeeze the two in one consistent response , which is doable in DNS terms , for the sake of the service ... or give shortened responses and iterate trough natural granularity of DNS messaging).Preemptive / ALL caching is less interesting to me personally , but I understand the relatively low effort required by infrastructure to take this extra burden of caching+refreshing it ( we now always desire to get the most records at once ... ) . Might state that too , that DNSSEC interactions will require more storage , or more bandwidth , or refer to it . Talking about Auth / Auth , OAuth or all the SSO you can name , are clearly BIG advantages ( i.e. refuse some clients or geo - locate / displace others ) . I was thinking MAC / DAC reasonings were to fall outside of DNS - over - HTTP draft , IMHO ... is not a "" native "" feature of DNS , as - is ... what is your opinion ? to me is HTTP.OTOH , a more precise Accounting is a native aspect coming straight out of doh ( making accounting easier if ever needed , my old implementation accounted for billing , not a new thing ) . To me , just because some companies do bill on imprecise Netflow stats , accurate tcpdumps on mirror ports , or even weirder things , does n't mean DOH can not state ( mention ) about "" billabilty "" at the net of SYN , SYN / ACK , ACK , retransmissions and RESETS :) :) Having a line like : "" as any other HTTP dialogue , DNS - over - HTTP might offer possibility to have its usage logged and billed "" . For some companies it might be a plus , let 's not tell them they could have done it with bind or plugins :) Traceability ? Is a mention to logging needed ? ( or needs to be implied as any HTTP does?).The detail of the query itself might not be available to authorities via such simplified logs anyway , and that has been made clear during the tables anyway , by regarding POSTs for example . In my previous exp in law enforcment for telco -started with landlines then switched to web- European ISP and hosting providers were obliged / required to provide HTTP and mail logs by investigating authorities ( at their own cost of maintenance , law enforcment might be a dead budget for smaller companies ) . It was 2006 , 2009 then 2016 things have n't much improved , just the sensibility towards the topic but not the smoothness of process and guarantee of security in personal / sensible data treatment , in medium companies not just . Many product offer since day 1 full auditing features , FGA in Oracle for example , simple or complex as be , is just matter of using them :( But why DNS traces are not always taken in consideration by all authorities , or everything becomes "" best - effort "" ? The queries / responses might not be traced / traceable , surely not in post - mortem , only some providers can / do . Not sponsoring any point of view on justice or privacy . ... this for doh is an evolution and an involution : the space needed to account for the long - term - storage might be an argument not to spouse doh . Obviously , filtering out is easy ( ct= ) By mentioning such aspect(s ) we can help answering frequent questions before they 're even been asked : +1 : As we are realising , mixing the experts from 2 different worlds of DNS and HTTP is not easy , so the more dialectical the best ? With clear examples for sure . I just think that people need to think what is DOH and what does and what does n't automagically do ( extra certificates , extra logs , "" slower "" but maybe faster , etc ) so to be able to compare pros / cons comparing to non HTTP / DNS servers / clients . To me is doh is just a transport thing , and I really like it since many years :) sorry for super long post , will copy this to mailing list if you think is useful . I agree with almost everything you say , but I do n't think the protocol doc is the place to got at all of these things at this length . I 'm just trying to give a flavor of the implications of using the http ecosystem - which you go into in much deeper detail . The document has to focus on how you do that . my suggestion would be you raise the concept of a separate applicability / use - case document if you think this is valuable .. you could raise this as part of the charter discussion for DoH. Or you could just work on a really great blog post :) Thank you I will follow all of your suggestions , just a matter of having some time to say something "" useful "" and hopefully , possibly correct . I am still willing to follow all of the discussions around the doh topic , I have a lot to learn from all of you , really proud of your consideration . I understand the logic of defining precise specifications and how - to - implement , proper of drafting RFC . I fully support the criteria of "" referring to "" other standing and specific RFCs by keeping the transversal implications outside of the doc , as it might over - confuse implementors and this is not the goal . Sorry for that , I just wanted to mention some of the possible "" arguments "" or side - effects ( my first time in participating in such exciting discussions and group ) . My congratulations to all of the working group and supporters , for all the efforts being put into this important draft" eb8e329701a1e22b3ee1e7e02f5e229649355f0f eb8e329701a1e22b3ee1e7e02f5e229649355f0f draft-ietf-doh-dns-over-https-latest 02fc2dc006a5e8528c3233c295ac4d2a2122d6052bf5e005ab5d6bfe3c3b1ff4 ['eaf86e75773cda3db897646dbb2845644406e315215006c451c16225d19eefb9', '02fc2dc006a5e8528c3233c295ac4d2a2122d6052bf5e005ab5d6bfe3c3b1ff4'] B "the internet does not always provide end to end reachability for native dns. on-path network devices may spoof dns responses, block dns requests, or just redirect dns queries to different dns servers that give less-than-honest answers. these are also sometimes delivered with poor performance or reduced feature sets." 99 {9, 46} {9: 0, 46: 1} {99: 0, 86: 1} "The described approach is more than a tunnel over HTTP. It establishes default media formatting types for requests and responses but uses normal HTTP content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. In addition to this media type negotiation, it aligns itself with HTTP features such as caching, redirection, proxying, and compression." "on list points out that we do n't mention some pretty key architectural benefits of doh such as authentication , amplification resistance , etc . on list as "" changes for draft - ietf - doh - dns - over - https-03 "" . is meant to address this Hi , following few impressions on advantageous aspects of doh : For sure , the experience aquired in mitigation against hammering attacks on "" HTTP "" layer is much more effective and well - ruled than DDoS mitigation(s ) when comparing to UDP DNS , due to persisting amplification attacks . Many different APIs to expire contents are available on a system layer ( Cloudflare , squid , find / rm ) , many HTTP request headers too on a protocl layer , so shall be widely possible to "" control "" cache validity with non - DNS methods , when system is responsive ( i.e. can take little longer).Compression , modernised with brotli and binary proto . I also sustain the need of a mention to the HEAD method as cache priming or verification method .. Is it desired , interesting , considered any useful by the working group ? ( Stephane brought this up to as a bug , but me too I was considering it since ever)Leveraging CDN caches ( when atomic , i.e. , CF - ID ) is another immediate advantage , but as discussed in the CNAME / A thread there might arise issues with ADDITIONAL / AUTHORITY sections ' validity ( my implementation used to squeeze the two in one consistent response , which is doable in DNS terms , for the sake of the service ... or give shortened responses and iterate trough natural granularity of DNS messaging).Preemptive / ALL caching is less interesting to me personally , but I understand the relatively low effort required by infrastructure to take this extra burden of caching+refreshing it ( we now always desire to get the most records at once ... ) . Might state that too , that DNSSEC interactions will require more storage , or more bandwidth , or refer to it . Talking about Auth / Auth , OAuth or all the SSO you can name , are clearly BIG advantages ( i.e. refuse some clients or geo - locate / displace others ) . I was thinking MAC / DAC reasonings were to fall outside of DNS - over - HTTP draft , IMHO ... is not a "" native "" feature of DNS , as - is ... what is your opinion ? to me is HTTP.OTOH , a more precise Accounting is a native aspect coming straight out of doh ( making accounting easier if ever needed , my old implementation accounted for billing , not a new thing ) . To me , just because some companies do bill on imprecise Netflow stats , accurate tcpdumps on mirror ports , or even weirder things , does n't mean DOH can not state ( mention ) about "" billabilty "" at the net of SYN , SYN / ACK , ACK , retransmissions and RESETS :) :) Having a line like : "" as any other HTTP dialogue , DNS - over - HTTP might offer possibility to have its usage logged and billed "" . For some companies it might be a plus , let 's not tell them they could have done it with bind or plugins :) Traceability ? Is a mention to logging needed ? ( or needs to be implied as any HTTP does?).The detail of the query itself might not be available to authorities via such simplified logs anyway , and that has been made clear during the tables anyway , by regarding POSTs for example . In my previous exp in law enforcment for telco -started with landlines then switched to web- European ISP and hosting providers were obliged / required to provide HTTP and mail logs by investigating authorities ( at their own cost of maintenance , law enforcment might be a dead budget for smaller companies ) . It was 2006 , 2009 then 2016 things have n't much improved , just the sensibility towards the topic but not the smoothness of process and guarantee of security in personal / sensible data treatment , in medium companies not just . Many product offer since day 1 full auditing features , FGA in Oracle for example , simple or complex as be , is just matter of using them :( But why DNS traces are not always taken in consideration by all authorities , or everything becomes "" best - effort "" ? The queries / responses might not be traced / traceable , surely not in post - mortem , only some providers can / do . Not sponsoring any point of view on justice or privacy . ... this for doh is an evolution and an involution : the space needed to account for the long - term - storage might be an argument not to spouse doh . Obviously , filtering out is easy ( ct= ) By mentioning such aspect(s ) we can help answering frequent questions before they 're even been asked : +1 : As we are realising , mixing the experts from 2 different worlds of DNS and HTTP is not easy , so the more dialectical the best ? With clear examples for sure . I just think that people need to think what is DOH and what does and what does n't automagically do ( extra certificates , extra logs , "" slower "" but maybe faster , etc ) so to be able to compare pros / cons comparing to non HTTP / DNS servers / clients . To me is doh is just a transport thing , and I really like it since many years :) sorry for super long post , will copy this to mailing list if you think is useful . I agree with almost everything you say , but I do n't think the protocol doc is the place to got at all of these things at this length . I 'm just trying to give a flavor of the implications of using the http ecosystem - which you go into in much deeper detail . The document has to focus on how you do that . my suggestion would be you raise the concept of a separate applicability / use - case document if you think this is valuable .. you could raise this as part of the charter discussion for DoH. Or you could just work on a really great blog post :) Thank you I will follow all of your suggestions , just a matter of having some time to say something "" useful "" and hopefully , possibly correct . I am still willing to follow all of the discussions around the doh topic , I have a lot to learn from all of you , really proud of your consideration . I understand the logic of defining precise specifications and how - to - implement , proper of drafting RFC . I fully support the criteria of "" referring to "" other standing and specific RFCs by keeping the transversal implications outside of the doc , as it might over - confuse implementors and this is not the goal . Sorry for that , I just wanted to mention some of the possible "" arguments "" or side - effects ( my first time in participating in such exciting discussions and group ) . My congratulations to all of the working group and supporters , for all the efforts being put into this important draft" eb8e329701a1e22b3ee1e7e02f5e229649355f0f eb8e329701a1e22b3ee1e7e02f5e229649355f0f draft-ietf-doh-dns-over-https-latest 0f36abea43127d9f707a7f4a944b9a878decf0eda32300fb5d31934569d91e19 ['0f36abea43127d9f707a7f4a944b9a878decf0eda32300fb5d31934569d91e19', '9efb9614a59ef5b1bf16e384a7b58c3ffce76db6c742f0a0d032d2c08fbc832f'] A "the described approach is more than a tunnel over http. it establishes default media formatting types for requests and responses but uses normal http content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. in addition to this media type negotiation, it aligns itself with http features such as caching, redirection, proxying, and compression." 99 {9, 46} {9: 0, 46: 1} {99: 0, 86: 1} "The described approach is more than a tunnel over HTTP. It establishes default media formatting types for requests and responses but uses normal HTTP content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. In addition to this media type negotiation, it aligns itself with HTTP features such as caching, redirection, proxying, authentication, and compression." "on list points out that we do n't mention some pretty key architectural benefits of doh such as authentication , amplification resistance , etc . on list as "" changes for draft - ietf - doh - dns - over - https-03 "" . is meant to address this Hi , following few impressions on advantageous aspects of doh : For sure , the experience aquired in mitigation against hammering attacks on "" HTTP "" layer is much more effective and well - ruled than DDoS mitigation(s ) when comparing to UDP DNS , due to persisting amplification attacks . Many different APIs to expire contents are available on a system layer ( Cloudflare , squid , find / rm ) , many HTTP request headers too on a protocl layer , so shall be widely possible to "" control "" cache validity with non - DNS methods , when system is responsive ( i.e. can take little longer).Compression , modernised with brotli and binary proto . I also sustain the need of a mention to the HEAD method as cache priming or verification method .. Is it desired , interesting , considered any useful by the working group ? ( Stephane brought this up to as a bug , but me too I was considering it since ever)Leveraging CDN caches ( when atomic , i.e. , CF - ID ) is another immediate advantage , but as discussed in the CNAME / A thread there might arise issues with ADDITIONAL / AUTHORITY sections ' validity ( my implementation used to squeeze the two in one consistent response , which is doable in DNS terms , for the sake of the service ... or give shortened responses and iterate trough natural granularity of DNS messaging).Preemptive / ALL caching is less interesting to me personally , but I understand the relatively low effort required by infrastructure to take this extra burden of caching+refreshing it ( we now always desire to get the most records at once ... ) . Might state that too , that DNSSEC interactions will require more storage , or more bandwidth , or refer to it . Talking about Auth / Auth , OAuth or all the SSO you can name , are clearly BIG advantages ( i.e. refuse some clients or geo - locate / displace others ) . I was thinking MAC / DAC reasonings were to fall outside of DNS - over - HTTP draft , IMHO ... is not a "" native "" feature of DNS , as - is ... what is your opinion ? to me is HTTP.OTOH , a more precise Accounting is a native aspect coming straight out of doh ( making accounting easier if ever needed , my old implementation accounted for billing , not a new thing ) . To me , just because some companies do bill on imprecise Netflow stats , accurate tcpdumps on mirror ports , or even weirder things , does n't mean DOH can not state ( mention ) about "" billabilty "" at the net of SYN , SYN / ACK , ACK , retransmissions and RESETS :) :) Having a line like : "" as any other HTTP dialogue , DNS - over - HTTP might offer possibility to have its usage logged and billed "" . For some companies it might be a plus , let 's not tell them they could have done it with bind or plugins :) Traceability ? Is a mention to logging needed ? ( or needs to be implied as any HTTP does?).The detail of the query itself might not be available to authorities via such simplified logs anyway , and that has been made clear during the tables anyway , by regarding POSTs for example . In my previous exp in law enforcment for telco -started with landlines then switched to web- European ISP and hosting providers were obliged / required to provide HTTP and mail logs by investigating authorities ( at their own cost of maintenance , law enforcment might be a dead budget for smaller companies ) . It was 2006 , 2009 then 2016 things have n't much improved , just the sensibility towards the topic but not the smoothness of process and guarantee of security in personal / sensible data treatment , in medium companies not just . Many product offer since day 1 full auditing features , FGA in Oracle for example , simple or complex as be , is just matter of using them :( But why DNS traces are not always taken in consideration by all authorities , or everything becomes "" best - effort "" ? The queries / responses might not be traced / traceable , surely not in post - mortem , only some providers can / do . Not sponsoring any point of view on justice or privacy . ... this for doh is an evolution and an involution : the space needed to account for the long - term - storage might be an argument not to spouse doh . Obviously , filtering out is easy ( ct= ) By mentioning such aspect(s ) we can help answering frequent questions before they 're even been asked : +1 : As we are realising , mixing the experts from 2 different worlds of DNS and HTTP is not easy , so the more dialectical the best ? With clear examples for sure . I just think that people need to think what is DOH and what does and what does n't automagically do ( extra certificates , extra logs , "" slower "" but maybe faster , etc ) so to be able to compare pros / cons comparing to non HTTP / DNS servers / clients . To me is doh is just a transport thing , and I really like it since many years :) sorry for super long post , will copy this to mailing list if you think is useful . I agree with almost everything you say , but I do n't think the protocol doc is the place to got at all of these things at this length . I 'm just trying to give a flavor of the implications of using the http ecosystem - which you go into in much deeper detail . The document has to focus on how you do that . my suggestion would be you raise the concept of a separate applicability / use - case document if you think this is valuable .. you could raise this as part of the charter discussion for DoH. Or you could just work on a really great blog post :) Thank you I will follow all of your suggestions , just a matter of having some time to say something "" useful "" and hopefully , possibly correct . I am still willing to follow all of the discussions around the doh topic , I have a lot to learn from all of you , really proud of your consideration . I understand the logic of defining precise specifications and how - to - implement , proper of drafting RFC . I fully support the criteria of "" referring to "" other standing and specific RFCs by keeping the transversal implications outside of the doc , as it might over - confuse implementors and this is not the goal . Sorry for that , I just wanted to mention some of the possible "" arguments "" or side - effects ( my first time in participating in such exciting discussions and group ) . My congratulations to all of the working group and supporters , for all the efforts being put into this important draft" eb8e329701a1e22b3ee1e7e02f5e229649355f0f eb8e329701a1e22b3ee1e7e02f5e229649355f0f draft-ietf-doh-dns-over-https-latest 9efb9614a59ef5b1bf16e384a7b58c3ffce76db6c742f0a0d032d2c08fbc832f ['0f36abea43127d9f707a7f4a944b9a878decf0eda32300fb5d31934569d91e19', '9efb9614a59ef5b1bf16e384a7b58c3ffce76db6c742f0a0d032d2c08fbc832f'] B "the described approach is more than a tunnel over http. it establishes default media formatting types for requests and responses but uses normal http content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. in addition to this media type negotiation, it aligns itself with http features such as caching, redirection, proxying, authentication, and compression." 99 {8, 68, 69, 7} {8: 0, 7: 1, 68: 2, 69: 2} {99: 0, 34: 1, 1: 2, 36: 2} "Running DNS over HTTPS relies on the security of the underlying HTTP transport. Implementations utilizing HTTP/2 benefit from the TLS profile defined in RFC7540 Section 9.2." "on list points out that we do n't mention some pretty key architectural benefits of doh such as authentication , amplification resistance , etc . on list as "" changes for draft - ietf - doh - dns - over - https-03 "" . is meant to address this Hi , following few impressions on advantageous aspects of doh : For sure , the experience aquired in mitigation against hammering attacks on "" HTTP "" layer is much more effective and well - ruled than DDoS mitigation(s ) when comparing to UDP DNS , due to persisting amplification attacks . Many different APIs to expire contents are available on a system layer ( Cloudflare , squid , find / rm ) , many HTTP request headers too on a protocl layer , so shall be widely possible to "" control "" cache validity with non - DNS methods , when system is responsive ( i.e. can take little longer).Compression , modernised with brotli and binary proto . I also sustain the need of a mention to the HEAD method as cache priming or verification method .. Is it desired , interesting , considered any useful by the working group ? ( Stephane brought this up to as a bug , but me too I was considering it since ever)Leveraging CDN caches ( when atomic , i.e. , CF - ID ) is another immediate advantage , but as discussed in the CNAME / A thread there might arise issues with ADDITIONAL / AUTHORITY sections ' validity ( my implementation used to squeeze the two in one consistent response , which is doable in DNS terms , for the sake of the service ... or give shortened responses and iterate trough natural granularity of DNS messaging).Preemptive / ALL caching is less interesting to me personally , but I understand the relatively low effort required by infrastructure to take this extra burden of caching+refreshing it ( we now always desire to get the most records at once ... ) . Might state that too , that DNSSEC interactions will require more storage , or more bandwidth , or refer to it . Talking about Auth / Auth , OAuth or all the SSO you can name , are clearly BIG advantages ( i.e. refuse some clients or geo - locate / displace others ) . I was thinking MAC / DAC reasonings were to fall outside of DNS - over - HTTP draft , IMHO ... is not a "" native "" feature of DNS , as - is ... what is your opinion ? to me is HTTP.OTOH , a more precise Accounting is a native aspect coming straight out of doh ( making accounting easier if ever needed , my old implementation accounted for billing , not a new thing ) . To me , just because some companies do bill on imprecise Netflow stats , accurate tcpdumps on mirror ports , or even weirder things , does n't mean DOH can not state ( mention ) about "" billabilty "" at the net of SYN , SYN / ACK , ACK , retransmissions and RESETS :) :) Having a line like : "" as any other HTTP dialogue , DNS - over - HTTP might offer possibility to have its usage logged and billed "" . For some companies it might be a plus , let 's not tell them they could have done it with bind or plugins :) Traceability ? Is a mention to logging needed ? ( or needs to be implied as any HTTP does?).The detail of the query itself might not be available to authorities via such simplified logs anyway , and that has been made clear during the tables anyway , by regarding POSTs for example . In my previous exp in law enforcment for telco -started with landlines then switched to web- European ISP and hosting providers were obliged / required to provide HTTP and mail logs by investigating authorities ( at their own cost of maintenance , law enforcment might be a dead budget for smaller companies ) . It was 2006 , 2009 then 2016 things have n't much improved , just the sensibility towards the topic but not the smoothness of process and guarantee of security in personal / sensible data treatment , in medium companies not just . Many product offer since day 1 full auditing features , FGA in Oracle for example , simple or complex as be , is just matter of using them :( But why DNS traces are not always taken in consideration by all authorities , or everything becomes "" best - effort "" ? The queries / responses might not be traced / traceable , surely not in post - mortem , only some providers can / do . Not sponsoring any point of view on justice or privacy . ... this for doh is an evolution and an involution : the space needed to account for the long - term - storage might be an argument not to spouse doh . Obviously , filtering out is easy ( ct= ) By mentioning such aspect(s ) we can help answering frequent questions before they 're even been asked : +1 : As we are realising , mixing the experts from 2 different worlds of DNS and HTTP is not easy , so the more dialectical the best ? With clear examples for sure . I just think that people need to think what is DOH and what does and what does n't automagically do ( extra certificates , extra logs , "" slower "" but maybe faster , etc ) so to be able to compare pros / cons comparing to non HTTP / DNS servers / clients . To me is doh is just a transport thing , and I really like it since many years :) sorry for super long post , will copy this to mailing list if you think is useful . I agree with almost everything you say , but I do n't think the protocol doc is the place to got at all of these things at this length . I 'm just trying to give a flavor of the implications of using the http ecosystem - which you go into in much deeper detail . The document has to focus on how you do that . my suggestion would be you raise the concept of a separate applicability / use - case document if you think this is valuable .. you could raise this as part of the charter discussion for DoH. Or you could just work on a really great blog post :) Thank you I will follow all of your suggestions , just a matter of having some time to say something "" useful "" and hopefully , possibly correct . I am still willing to follow all of the discussions around the doh topic , I have a lot to learn from all of you , really proud of your consideration . I understand the logic of defining precise specifications and how - to - implement , proper of drafting RFC . I fully support the criteria of "" referring to "" other standing and specific RFCs by keeping the transversal implications outside of the doc , as it might over - confuse implementors and this is not the goal . Sorry for that , I just wanted to mention some of the possible "" arguments "" or side - effects ( my first time in participating in such exciting discussions and group ) . My congratulations to all of the working group and supporters , for all the efforts being put into this important draft" eb8e329701a1e22b3ee1e7e02f5e229649355f0f eb8e329701a1e22b3ee1e7e02f5e229649355f0f draft-ietf-doh-dns-over-https-latest 0d3de8ab674456fbd0fb436cdb04209912d5fb844d8ff5a7b130d20fd46e024a ['0d3de8ab674456fbd0fb436cdb04209912d5fb844d8ff5a7b130d20fd46e024a', 'ba3c7eea84de95d6f8d01f47a2f0e2934d6963ea8a79136308331b9e5943a200'] A "running dns over https relies on the security of the underlying http transport. implementations utilizing http/2 benefit from the tls profile defined in rfc7540 section 9.2." 99 {8, 68, 69, 7} {8: 0, 7: 1, 68: 2, 69: 2} {99: 0, 34: 1, 1: 2, 36: 2} "Running DNS over HTTPS relies on the security of the underlying HTTP transport. This mitigates classic amplication attacks for UDP-based DNS. Implementations utilizing HTTP/2 benefit from the TLS profile defined in RFC7540 Section 9.2." "on list points out that we do n't mention some pretty key architectural benefits of doh such as authentication , amplification resistance , etc . on list as "" changes for draft - ietf - doh - dns - over - https-03 "" . is meant to address this Hi , following few impressions on advantageous aspects of doh : For sure , the experience aquired in mitigation against hammering attacks on "" HTTP "" layer is much more effective and well - ruled than DDoS mitigation(s ) when comparing to UDP DNS , due to persisting amplification attacks . Many different APIs to expire contents are available on a system layer ( Cloudflare , squid , find / rm ) , many HTTP request headers too on a protocl layer , so shall be widely possible to "" control "" cache validity with non - DNS methods , when system is responsive ( i.e. can take little longer).Compression , modernised with brotli and binary proto . I also sustain the need of a mention to the HEAD method as cache priming or verification method .. Is it desired , interesting , considered any useful by the working group ? ( Stephane brought this up to as a bug , but me too I was considering it since ever)Leveraging CDN caches ( when atomic , i.e. , CF - ID ) is another immediate advantage , but as discussed in the CNAME / A thread there might arise issues with ADDITIONAL / AUTHORITY sections ' validity ( my implementation used to squeeze the two in one consistent response , which is doable in DNS terms , for the sake of the service ... or give shortened responses and iterate trough natural granularity of DNS messaging).Preemptive / ALL caching is less interesting to me personally , but I understand the relatively low effort required by infrastructure to take this extra burden of caching+refreshing it ( we now always desire to get the most records at once ... ) . Might state that too , that DNSSEC interactions will require more storage , or more bandwidth , or refer to it . Talking about Auth / Auth , OAuth or all the SSO you can name , are clearly BIG advantages ( i.e. refuse some clients or geo - locate / displace others ) . I was thinking MAC / DAC reasonings were to fall outside of DNS - over - HTTP draft , IMHO ... is not a "" native "" feature of DNS , as - is ... what is your opinion ? to me is HTTP.OTOH , a more precise Accounting is a native aspect coming straight out of doh ( making accounting easier if ever needed , my old implementation accounted for billing , not a new thing ) . To me , just because some companies do bill on imprecise Netflow stats , accurate tcpdumps on mirror ports , or even weirder things , does n't mean DOH can not state ( mention ) about "" billabilty "" at the net of SYN , SYN / ACK , ACK , retransmissions and RESETS :) :) Having a line like : "" as any other HTTP dialogue , DNS - over - HTTP might offer possibility to have its usage logged and billed "" . For some companies it might be a plus , let 's not tell them they could have done it with bind or plugins :) Traceability ? Is a mention to logging needed ? ( or needs to be implied as any HTTP does?).The detail of the query itself might not be available to authorities via such simplified logs anyway , and that has been made clear during the tables anyway , by regarding POSTs for example . In my previous exp in law enforcment for telco -started with landlines then switched to web- European ISP and hosting providers were obliged / required to provide HTTP and mail logs by investigating authorities ( at their own cost of maintenance , law enforcment might be a dead budget for smaller companies ) . It was 2006 , 2009 then 2016 things have n't much improved , just the sensibility towards the topic but not the smoothness of process and guarantee of security in personal / sensible data treatment , in medium companies not just . Many product offer since day 1 full auditing features , FGA in Oracle for example , simple or complex as be , is just matter of using them :( But why DNS traces are not always taken in consideration by all authorities , or everything becomes "" best - effort "" ? The queries / responses might not be traced / traceable , surely not in post - mortem , only some providers can / do . Not sponsoring any point of view on justice or privacy . ... this for doh is an evolution and an involution : the space needed to account for the long - term - storage might be an argument not to spouse doh . Obviously , filtering out is easy ( ct= ) By mentioning such aspect(s ) we can help answering frequent questions before they 're even been asked : +1 : As we are realising , mixing the experts from 2 different worlds of DNS and HTTP is not easy , so the more dialectical the best ? With clear examples for sure . I just think that people need to think what is DOH and what does and what does n't automagically do ( extra certificates , extra logs , "" slower "" but maybe faster , etc ) so to be able to compare pros / cons comparing to non HTTP / DNS servers / clients . To me is doh is just a transport thing , and I really like it since many years :) sorry for super long post , will copy this to mailing list if you think is useful . I agree with almost everything you say , but I do n't think the protocol doc is the place to got at all of these things at this length . I 'm just trying to give a flavor of the implications of using the http ecosystem - which you go into in much deeper detail . The document has to focus on how you do that . my suggestion would be you raise the concept of a separate applicability / use - case document if you think this is valuable .. you could raise this as part of the charter discussion for DoH. Or you could just work on a really great blog post :) Thank you I will follow all of your suggestions , just a matter of having some time to say something "" useful "" and hopefully , possibly correct . I am still willing to follow all of the discussions around the doh topic , I have a lot to learn from all of you , really proud of your consideration . I understand the logic of defining precise specifications and how - to - implement , proper of drafting RFC . I fully support the criteria of "" referring to "" other standing and specific RFCs by keeping the transversal implications outside of the doc , as it might over - confuse implementors and this is not the goal . Sorry for that , I just wanted to mention some of the possible "" arguments "" or side - effects ( my first time in participating in such exciting discussions and group ) . My congratulations to all of the working group and supporters , for all the efforts being put into this important draft" eb8e329701a1e22b3ee1e7e02f5e229649355f0f eb8e329701a1e22b3ee1e7e02f5e229649355f0f draft-ietf-doh-dns-over-https-latest ba3c7eea84de95d6f8d01f47a2f0e2934d6963ea8a79136308331b9e5943a200 ['0d3de8ab674456fbd0fb436cdb04209912d5fb844d8ff5a7b130d20fd46e024a', 'ba3c7eea84de95d6f8d01f47a2f0e2934d6963ea8a79136308331b9e5943a200'] B "running dns over https relies on the security of the underlying http transport. this mitigates classic amplication attacks for udp-based dns. implementations utilizing http/2 benefit from the tls profile defined in rfc7540 section 9.2." 98 {} {} {} "Local policy considerations and similar factors mean different DNS servers may provide different results to the same query: for instance in split DNS configurations RFC6950. It logically follows that the server which is queried can influence the end result. Therefore a client's choice of DNS server may affect the responses it gets to its queries." "wrt mailing list thread and DOH andrew sullivan suggests modifying an existing paragraph in operational considerations to have one new sentence : servers may provide different results to the same query : for instance in split DNS configurations [ RFC6950 ] . It logically follows that the server which is queried can influence the end result . Therefore a client 's choice of DNS server may affect the responses it gets to its queries . In the case of DNS64 [ RFC6147 ] , the choice could affect whether IPv6 / IPv4 translation will work at all . in my opinion that 's fine . We do n't need to enumerate everything , but was already called out as a non goal so we might as well note this implication of that . PR" f6e5c766c7ee12cc6f26a2fe2df0ff177c962ae6 f6e5c766c7ee12cc6f26a2fe2df0ff177c962ae6 draft-ietf-doh-dns-over-https-latest 8604b65a05234482d9835691813b2b863ce2f58b4dfa6792056f2937ce4296d8 ['1ef036c5b4e949e33c26d1f15d4cda0540d87450e7f1bae83eb535a5cad4a1d1', '8604b65a05234482d9835691813b2b863ce2f58b4dfa6792056f2937ce4296d8'] A "local policy considerations and similar factors mean different dns servers may provide different results to the same query: for instance in split dns configurations rfc6950. it logically follows that the server which is queried can influence the end result. therefore a client's choice of dns server may affect the responses it gets to its queries." 98 {} {} {} "Local policy considerations and similar factors mean different DNS servers may provide different results to the same query: for instance in split DNS configurations RFC6950. It logically follows that the server which is queried can influence the end result. Therefore a client's choice of DNS server may affect the responses it gets to its queries. For example, in the case of DNS64 RFC6147, the choice could affect whether IPv6/IPv4 translation will work at all." "wrt mailing list thread and DOH andrew sullivan suggests modifying an existing paragraph in operational considerations to have one new sentence : servers may provide different results to the same query : for instance in split DNS configurations [ RFC6950 ] . It logically follows that the server which is queried can influence the end result . Therefore a client 's choice of DNS server may affect the responses it gets to its queries . In the case of DNS64 [ RFC6147 ] , the choice could affect whether IPv6 / IPv4 translation will work at all . in my opinion that 's fine . We do n't need to enumerate everything , but was already called out as a non goal so we might as well note this implication of that . PR" f6e5c766c7ee12cc6f26a2fe2df0ff177c962ae6 f6e5c766c7ee12cc6f26a2fe2df0ff177c962ae6 draft-ietf-doh-dns-over-https-latest 1ef036c5b4e949e33c26d1f15d4cda0540d87450e7f1bae83eb535a5cad4a1d1 ['1ef036c5b4e949e33c26d1f15d4cda0540d87450e7f1bae83eb535a5cad4a1d1', '8604b65a05234482d9835691813b2b863ce2f58b4dfa6792056f2937ce4296d8'] B "local policy considerations and similar factors mean different dns servers may provide different results to the same query: for instance in split dns configurations rfc6950. it logically follows that the server which is queried can influence the end result. therefore a client's choice of dns server may affect the responses it gets to its queries. for example, in the case of dns64 rfc6147, the choice could affect whether ipv6/ipv4 translation will work at all." 97 {116, 117, 63} {63: 0, 117: 1, 116: 2} {97: 0, 120: 1, 123: 2} The data payload is the DNS on-the-wire format defined in RFC1035. "Make clear here that it does n't intrinsically come with a content - length limit , despite traditional DNS message length limits ? I merged this in a race condition with 's comments . I support that change , but its a little different than the bug that drove this . can you just make the change directly and minimally in this same text ? it could use a preview to help the reader orient herself See" 0d77972aa4d4a243c2be74eb5aab4994c6b76ba6 0d77972aa4d4a243c2be74eb5aab4994c6b76ba6 draft-ietf-doh-dns-over-https-latest 89175952a6ab670628c2bbbc4a1205c044442d6260a96dcb3ae949ffd6f3e8c2 ['89175952a6ab670628c2bbbc4a1205c044442d6260a96dcb3ae949ffd6f3e8c2', 'fe9042f1dcc33ce908506467528fbe6c1eee30ed621b81b63c253c41218fc066'] A the data payload is the dns on-the-wire format defined in rfc1035. 97 {116, 117, 63} {63: 0, 117: 1, 116: 2} {97: 0, 120: 1, 123: 2} "The data payload is the DNS on-the-wire format defined in RFC1035. The format is for DNS over UDP. (Note that this is different than the wire format used in RFC7858." "Make clear here that it does n't intrinsically come with a content - length limit , despite traditional DNS message length limits ? I merged this in a race condition with 's comments . I support that change , but its a little different than the bug that drove this . can you just make the change directly and minimally in this same text ? it could use a preview to help the reader orient herself See" 0d77972aa4d4a243c2be74eb5aab4994c6b76ba6 0d77972aa4d4a243c2be74eb5aab4994c6b76ba6 draft-ietf-doh-dns-over-https-latest fe9042f1dcc33ce908506467528fbe6c1eee30ed621b81b63c253c41218fc066 ['89175952a6ab670628c2bbbc4a1205c044442d6260a96dcb3ae949ffd6f3e8c2', 'fe9042f1dcc33ce908506467528fbe6c1eee30ed621b81b63c253c41218fc066'] B "the data payload is the dns on-the-wire format defined in rfc1035. the format is for dns over udp. (note that this is different than the wire format used in rfc7858." 96 {} {} {} "The HTTPS connection provides transport security for the interaction between the DNS API server and client, but does not inherently ensure the authenticity of DNS data. A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server, much as a DNS client might choose to trust answers from its recursive DNS resolver." "… dnssec on list writes : Section 6 of -03 says this : Different response media types will provide more or less information from a DNS response . For example , one response type might include the information from the DNS header bytes while another might omit it . But section 9 says A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server , much as a DNS client might choose to trust answers from its recursive DNS resolver . It seems to me that these are in tension with one another , because the AD and CD bits are in the header that the response type is permitted to throw away . Maybe it could be resolved thus : NEW A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server , much as a DNS client might choose to trust answers from its recursive DNS resolver . This capability might be affected by the response media type a DNS API server supports ." e161fbc780a5b6f2f1988ec9ccff06cd15ca4ebc e161fbc780a5b6f2f1988ec9ccff06cd15ca4ebc draft-ietf-doh-dns-over-https-latest 305096016c9956c09a589148d652168d1d3ee56f7bdd1a472327d98fe5cf4d70 ['305096016c9956c09a589148d652168d1d3ee56f7bdd1a472327d98fe5cf4d70', '20077f6db1c0db91d43813a84ce13d39477527b70c7908b2bb2c931e47379096'] A "the https connection provides transport security for the interaction between the dns api server and client, but does not inherently ensure the authenticity of dns data. a dns api client may also perform full dnssec validation of answers received from a dns api server or it may choose to trust answers from a particular dns api server, much as a dns client might choose to trust answers from its recursive dns resolver." 96 {} {} {} "The HTTPS connection provides transport security for the interaction between the DNS API server and client, but does not inherently ensure the authenticity of DNS data. A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server, much as a DNS client might choose to trust answers from its recursive DNS resolver. This capability might be affected by the response media type." "… dnssec on list writes : Section 6 of -03 says this : Different response media types will provide more or less information from a DNS response . For example , one response type might include the information from the DNS header bytes while another might omit it . But section 9 says A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server , much as a DNS client might choose to trust answers from its recursive DNS resolver . It seems to me that these are in tension with one another , because the AD and CD bits are in the header that the response type is permitted to throw away . Maybe it could be resolved thus : NEW A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server , much as a DNS client might choose to trust answers from its recursive DNS resolver . This capability might be affected by the response media type a DNS API server supports ." e161fbc780a5b6f2f1988ec9ccff06cd15ca4ebc e161fbc780a5b6f2f1988ec9ccff06cd15ca4ebc draft-ietf-doh-dns-over-https-latest 20077f6db1c0db91d43813a84ce13d39477527b70c7908b2bb2c931e47379096 ['305096016c9956c09a589148d652168d1d3ee56f7bdd1a472327d98fe5cf4d70', '20077f6db1c0db91d43813a84ce13d39477527b70c7908b2bb2c931e47379096'] B "the https connection provides transport security for the interaction between the dns api server and client, but does not inherently ensure the authenticity of dns data. a dns api client may also perform full dnssec validation of answers received from a dns api server or it may choose to trust answers from a particular dns api server, much as a dns client might choose to trust answers from its recursive dns resolver. this capability might be affected by the response media type." 92 {41, 97} {97: 0, 41: 1} {92: 0, 86: 1} "An HTTP response with a 2xx status code (RFC7231 Section 6.3) indicates a valid DNS response to the query made in the HTTP request. A valid DNS response includes both success and failure responses." "intended largely a restating of mailing list thread : A question on the mix of DNS and HTTP semantics keep pr but reopen the issue to add a little more text PRs during meetings iz hArd" 55ed1207627ff9431b7e92ba32b511b8408a0980 55ed1207627ff9431b7e92ba32b511b8408a0980 draft-ietf-doh-dns-over-https-latest 65f78336ab65de55f280610452578dfef4ce82eae5958e9f44cffcd9211022a3 ['cdbe10f671d986340d5157bdc85dd7356057fe056df5ef33e78c8911d3cfb38b', '65f78336ab65de55f280610452578dfef4ce82eae5958e9f44cffcd9211022a3'] A "an http response with a 2xx status code (rfc7231 section 6.3) indicates a valid dns response to the query made in the http request. a valid dns response includes both success and failure responses." 92 {41, 97} {97: 0, 41: 1} {92: 0, 86: 1} "An HTTP response with a 2xx status code (RFC7231 Section 6.3) indicates a valid DNS response to the query made in the HTTP request. A valid DNS response includes both success and failure responses. For example, a DNS failure response such as SERVFAIL or NXDOMAIN will contain a successful HTTP 2xx status code even though there was a failure at the DNS layer. Responses with non-successful HTTP status codes do not contain DNS answers to the question in the corresponding request. Some of these non-successful HTTPS responses (such as redirects or authentication failures) could allow clients to make new requests to satisfy the original question." "intended largely a restating of mailing list thread : A question on the mix of DNS and HTTP semantics keep pr but reopen the issue to add a little more text PRs during meetings iz hArd" 55ed1207627ff9431b7e92ba32b511b8408a0980 55ed1207627ff9431b7e92ba32b511b8408a0980 draft-ietf-doh-dns-over-https-latest cdbe10f671d986340d5157bdc85dd7356057fe056df5ef33e78c8911d3cfb38b ['cdbe10f671d986340d5157bdc85dd7356057fe056df5ef33e78c8911d3cfb38b', '65f78336ab65de55f280610452578dfef4ce82eae5958e9f44cffcd9211022a3'] B "an http response with a 2xx status code (rfc7231 section 6.3) indicates a valid dns response to the query made in the http request. a valid dns response includes both success and failure responses. for example, a dns failure response such as servfail or nxdomain will contain a successful http 2xx status code even though there was a failure at the dns layer. responses with non-successful http status codes do not contain dns answers to the question in the corresponding request. some of these non-successful https responses (such as redirects or authentication failures) could allow clients to make new requests to satisfy the original question." 91 {16, 11} {16: 0, 11: 1} {91: 0, 173: 1} "Session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. Sections 10.6 (Compression) and 10.7 (Padding) of RFC7540 provide some further advice on mitigations within an HTTP/2 context." "In section 9 of DOH , it makes the following references : This is n't one of the formats that the tools process will convert into deep links . Instead , the "" 10.6 "" becomes a link to section 10.6 of DOH ( which does n't exist ) , while 10.7 does n't become a link at all . thanks . ugh . HTTP/2 provides further advice about the use of compression ( see Section 10.6 of [ RFC7540 ] ) and padding ( Section 10.7 of [ RFC7540 ] ) . That removes the "" mitigations "" word . Fixed in PR" 36ed29122ee9811e20bda8f2491913e28e30a3b3 36ed29122ee9811e20bda8f2491913e28e30a3b3 draft-ietf-doh-dns-over-https-latest 1a965388eff19871565a9e782fd7e488336bd479e9d385b7a8cc33963542246d ['1a965388eff19871565a9e782fd7e488336bd479e9d385b7a8cc33963542246d', '11057a9419a7e65c6b23f3135e3f7154dce0d6decbddac19d19dafa5b05f4f61'] A "session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. sections 10.6 (compression) and 10.7 (padding) of rfc7540 provide some further advice on mitigations within an http/2 context." 91 {16, 11} {16: 0, 11: 1} {91: 0, 173: 1} "Session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with DNS queries. HTTP/2 provides further advice about the use of compression (Section 10.6 of RFC7540) and padding (Section 10.7 of RFC7540)." "In section 9 of DOH , it makes the following references : This is n't one of the formats that the tools process will convert into deep links . Instead , the "" 10.6 "" becomes a link to section 10.6 of DOH ( which does n't exist ) , while 10.7 does n't become a link at all . thanks . ugh . HTTP/2 provides further advice about the use of compression ( see Section 10.6 of [ RFC7540 ] ) and padding ( Section 10.7 of [ RFC7540 ] ) . That removes the "" mitigations "" word . Fixed in PR" 36ed29122ee9811e20bda8f2491913e28e30a3b3 36ed29122ee9811e20bda8f2491913e28e30a3b3 draft-ietf-doh-dns-over-https-latest 11057a9419a7e65c6b23f3135e3f7154dce0d6decbddac19d19dafa5b05f4f61 ['1a965388eff19871565a9e782fd7e488336bd479e9d385b7a8cc33963542246d', '11057a9419a7e65c6b23f3135e3f7154dce0d6decbddac19d19dafa5b05f4f61'] B "session level encryption has well known weaknesses with respect to traffic analysis which might be particularly acute when dealing with dns queries. http/2 provides further advice about the use of compression (section 10.6 of rfc7540) and padding (section 10.7 of rfc7540)." 89 {49, 84, 85} {49: 0, 85: 1, 84: 2} {89: 0, 116: 1, 187: 2} "A client can use DNS over HTTPS as one of multiple mechanisms to obtain DNS data. If a client of this protocol encounters an HTTP error after sending a DNS query, and then falls back to a different DNS retrieval mechanism, doing so can weaken the privacy expected by the user of the client." "If you 're ok with the language I think this is ready to go into the draft based on list discussion . ( this does imply to has consensus - everything needs consensus , but as a process matter we need to try and channel those discussions into text for consideration .. its open to feedback , of course ) forked off from which deals with error handling , text on what happens if you choose to handle errors in a non doh fashion related to mailing list thread : A question on the mix of DNS and HTTP semantics" 95f01716425644305cf9898401c0464a5aaec5ab 95f01716425644305cf9898401c0464a5aaec5ab draft-ietf-doh-dns-over-https-latest 72d1aa8c65389efe1cf83588d7d4893742505d4335b4565fac9e7256374d7d8e ['72d1aa8c65389efe1cf83588d7d4893742505d4335b4565fac9e7256374d7d8e'] B "a client can use dns over https as one of multiple mechanisms to obtain dns data. if a client of this protocol encounters an http error after sending a dns query, and then falls back to a different dns retrieval mechanism, doing so can weaken the privacy expected by the user of the client." 86 {9, 46} {46: 0, 9: 1} {86: 0, 99: 1} "The described approach is more than a tunnel over HTTP. It establishes default media formatting types for requests and responses but uses normal HTTP content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. In addition to this media type negotiation, it aligns itself with HTTP features such as caching, proxying, and compression." "That s better . Thanks It would be good to explicitly call out the DNS rcode SERVFAIL as a type of failure response as well , not just NXDOMAIN , as they are very different sorts of failures such that the latter can be considered , in some contexts , as not really a failure . the merged patch no longer mentions specific rcodes at all .. wdyt ? I personally think that as a DNS implementer that there is enough ambiguity around understanding a valid DNS response , a negative answer , and a failure response , that it should be made more painfully clear that something like all DNS servers returning message - valid SERVFAILs still warrants an HTTP 2xx . Agree with . will reopen the issue to add more text largely a restating of mailing list thread : A question on the mix of DNS and HTTP semantics keep pr but reopen the issue to add a little more text rewording for the response text , but removed the security considerations until further discussion . , please edit or pull ." 8be369461dfed20e5d5baf0f08ac581df2b7848c 8be369461dfed20e5d5baf0f08ac581df2b7848c draft-ietf-doh-dns-over-https-latest 6e00770e2df0dcbdfcb809a4d60da47f997922bd52e139ff38038c36f6ae130c ['6e00770e2df0dcbdfcb809a4d60da47f997922bd52e139ff38038c36f6ae130c', '0f36abea43127d9f707a7f4a944b9a878decf0eda32300fb5d31934569d91e19'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/86/files A "the described approach is more than a tunnel over http. it establishes default media formatting types for requests and responses but uses normal http content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. in addition to this media type negotiation, it aligns itself with http features such as caching, proxying, and compression." 86 {9, 46} {46: 0, 9: 1} {86: 0, 99: 1} "The described approach is more than a tunnel over HTTP. It establishes default media formatting types for requests and responses but uses normal HTTP content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. In addition to this media type negotiation, it aligns itself with HTTP features such as caching, redirection, proxying, and compression." "That s better . Thanks It would be good to explicitly call out the DNS rcode SERVFAIL as a type of failure response as well , not just NXDOMAIN , as they are very different sorts of failures such that the latter can be considered , in some contexts , as not really a failure . the merged patch no longer mentions specific rcodes at all .. wdyt ? I personally think that as a DNS implementer that there is enough ambiguity around understanding a valid DNS response , a negative answer , and a failure response , that it should be made more painfully clear that something like all DNS servers returning message - valid SERVFAILs still warrants an HTTP 2xx . Agree with . will reopen the issue to add more text largely a restating of mailing list thread : A question on the mix of DNS and HTTP semantics keep pr but reopen the issue to add a little more text rewording for the response text , but removed the security considerations until further discussion . , please edit or pull ." 8be369461dfed20e5d5baf0f08ac581df2b7848c 8be369461dfed20e5d5baf0f08ac581df2b7848c draft-ietf-doh-dns-over-https-latest 0f36abea43127d9f707a7f4a944b9a878decf0eda32300fb5d31934569d91e19 ['6e00770e2df0dcbdfcb809a4d60da47f997922bd52e139ff38038c36f6ae130c', '0f36abea43127d9f707a7f4a944b9a878decf0eda32300fb5d31934569d91e19'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/86/files B "the described approach is more than a tunnel over http. it establishes default media formatting types for requests and responses but uses normal http content negotiation mechanisms for selecting alternatives that endpoints may prefer in anticipation of serving new use cases. in addition to this media type negotiation, it aligns itself with http features such as caching, redirection, proxying, and compression." 86 {41, 97} {41: 0, 97: 1} {86: 0, 92: 1} "An HTTP response with a 2xx status code (RFC7231 Section 6.3) indicates a valid DNS response to the query made in the HTTP request. A valid DNS response includes both success and failure responses." "That s better . Thanks It would be good to explicitly call out the DNS rcode SERVFAIL as a type of failure response as well , not just NXDOMAIN , as they are very different sorts of failures such that the latter can be considered , in some contexts , as not really a failure . the merged patch no longer mentions specific rcodes at all .. wdyt ? I personally think that as a DNS implementer that there is enough ambiguity around understanding a valid DNS response , a negative answer , and a failure response , that it should be made more painfully clear that something like all DNS servers returning message - valid SERVFAILs still warrants an HTTP 2xx . Agree with . will reopen the issue to add more text largely a restating of mailing list thread : A question on the mix of DNS and HTTP semantics keep pr but reopen the issue to add a little more text rewording for the response text , but removed the security considerations until further discussion . , please edit or pull ." 8be369461dfed20e5d5baf0f08ac581df2b7848c 8be369461dfed20e5d5baf0f08ac581df2b7848c draft-ietf-doh-dns-over-https-latest 65f78336ab65de55f280610452578dfef4ce82eae5958e9f44cffcd9211022a3 ['65f78336ab65de55f280610452578dfef4ce82eae5958e9f44cffcd9211022a3'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/86/files B "an http response with a 2xx status code (rfc7231 section 6.3) indicates a valid dns response to the query made in the http request. a valid dns response includes both success and failure responses." 84 {} {} {} "HTTP RFC7230 is a stateless application level protocol and therefore DOH implementations do not provide stateful ordering guarantees between different requests. DOH cannot be used as a transport for other protocols that require strict ordering." we can provide a reminder that http is stateless and therefore reordering happens so do n't use doh for a session protocol . d57e6cb54b30959a8cf386af5dc92e91665affe0 d57e6cb54b30959a8cf386af5dc92e91665affe0 draft-ietf-doh-dns-over-https-latest 0849ec27c6374ab8a7e1b23c50727e386c0fff4aeb337127a869836807f90edf ['0849ec27c6374ab8a7e1b23c50727e386c0fff4aeb337127a869836807f90edf'] B "http rfc7230 is a stateless application level protocol and therefore doh implementations do not provide stateful ordering guarantees between different requests. doh cannot be used as a transport for other protocols that require strict ordering." 81 {} {} {} "DNS queries sometimes experience problems with end to end connectivity at times and places where HTTPS flows freely." "... says : "" "" "" DNS queries sometimes experience problems with end to end connectivity at times and places where HTTPS flows freely . HTTPS provides the most practical mechanism for reliable end to end communication . Its use of TLS provides integrity and confidentiality guarantees and its use of HTTP allows it to interoperate with proxies , firewalls , and authentication systems where required for transit . "" "" "" Do we need to be talking about the motivation here ? The way it 's written sounds like it 's using HTTP for tunnelling , which is something we wanted to avoid ( especially if future efforts take away the message that HTTPS is the best tunnel ) . I think the introduction does a decent job of dealing with the "" not just a tunnel "" question .. and therefore these motivators can live there and we can have a very simple 1 line abstract . ( cutting non normative text is in vogue ) .. so I 'll put up a PR that reflects that ." 14445eb3fa21a10e2ac7f94110b352576977d85e 14445eb3fa21a10e2ac7f94110b352576977d85e draft-ietf-doh-dns-over-https-latest 0b841558118b456627b2ca3ba747543690566edc4afe856fb3d9b5f2a6ff2127 ['42e1f18ceaeb2204185ac4f001d02ae30355b15c01632d69157e4639605d84f7', '0b841558118b456627b2ca3ba747543690566edc4afe856fb3d9b5f2a6ff2127'] A "dns queries sometimes experience problems with end to end connectivity at times and places where https flows freely." 81 {} {} {} "HTTPS provides the most practical mechanism for reliable end to end communication. Its use of TLS provides integrity and confidentiality guarantees and its use of HTTP allows it to interoperate with proxies, firewalls, and authentication systems where required for transit." "... says : "" "" "" DNS queries sometimes experience problems with end to end connectivity at times and places where HTTPS flows freely . HTTPS provides the most practical mechanism for reliable end to end communication . Its use of TLS provides integrity and confidentiality guarantees and its use of HTTP allows it to interoperate with proxies , firewalls , and authentication systems where required for transit . "" "" "" Do we need to be talking about the motivation here ? The way it 's written sounds like it 's using HTTP for tunnelling , which is something we wanted to avoid ( especially if future efforts take away the message that HTTPS is the best tunnel ) . I think the introduction does a decent job of dealing with the "" not just a tunnel "" question .. and therefore these motivators can live there and we can have a very simple 1 line abstract . ( cutting non normative text is in vogue ) .. so I 'll put up a PR that reflects that ." 14445eb3fa21a10e2ac7f94110b352576977d85e 14445eb3fa21a10e2ac7f94110b352576977d85e draft-ietf-doh-dns-over-https-latest 42e1f18ceaeb2204185ac4f001d02ae30355b15c01632d69157e4639605d84f7 ['42e1f18ceaeb2204185ac4f001d02ae30355b15c01632d69157e4639605d84f7', '0b841558118b456627b2ca3ba747543690566edc4afe856fb3d9b5f2a6ff2127'] A "https provides the most practical mechanism for reliable end to end communication. its use of tls provides integrity and confidentiality guarantees and its use of http allows it to interoperate with proxies, firewalls, and authentication systems where required for transit." 80 {96, 22, 23, 58, 91} {91: 0, 96: 1, 22: 2, 23: 2, 58: 3} {80: 0, 64: 1, 60: 2, 43: 2, 19: 3} "When using the GET method the URI path MUST contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the dns parameter. The value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&dns=...) or it may be empty. An empty value indicates the default application/dns- udpwireformat type (e.g. ct&dns=...)." "… ry parameter The spec refers to "" query parameters "" , but does not specify their syntax . The most current reference for its definition is : If you take , then you conveniently avoid having to normatively reference that specification :)" 95ada91d047deefd694a82cf87c801add0d65a7c 95ada91d047deefd694a82cf87c801add0d65a7c draft-ietf-doh-dns-over-https-latest e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d ['c70e3d41073928d11a63e1d6a3cc9c92a502f60fdba44e34a2f2cd1892949cb1', 'e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d'] https://url.spec.whatwg.org/#application/x-www-form-urlencoded A "when using the get method the uri path must contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the dns parameter. the value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&dns=...) or it may be empty. an empty value indicates the default application/dns- udpwireformat type (e.g. ct&dns=...)." 80 {96, 22, 23, 58, 91} {91: 0, 96: 1, 22: 2, 23: 2, 58: 3} {80: 0, 64: 1, 60: 2, 43: 2, 19: 3} "When using the GET method the URI path MUST contain a query parameter name-value pair QUERYPARAMETER with the name of ""ct"" and a value indicating the media-format used for the dns parameter. The value may either be an explicit media type (e.g. ct=application/dns- udpwireformat&dns=...) or it may be empty. An empty value indicates the default application/dns-udpwireformat type (e.g. ct&dns=...)." "… ry parameter The spec refers to "" query parameters "" , but does not specify their syntax . The most current reference for its definition is : If you take , then you conveniently avoid having to normatively reference that specification :)" 95ada91d047deefd694a82cf87c801add0d65a7c 95ada91d047deefd694a82cf87c801add0d65a7c draft-ietf-doh-dns-over-https-latest c70e3d41073928d11a63e1d6a3cc9c92a502f60fdba44e34a2f2cd1892949cb1 ['c70e3d41073928d11a63e1d6a3cc9c92a502f60fdba44e34a2f2cd1892949cb1', 'e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d'] https://url.spec.whatwg.org/#application/x-www-form-urlencoded B "when using the get method the uri path must contain a query parameter name-value pair queryparameter with the name of ""ct"" and a value indicating the media-format used for the dns parameter. the value may either be an explicit media type (e.g. ct=application/dns- udpwireformat&dns=...) or it may be empty. an empty value indicates the default application/dns-udpwireformat type (e.g. ct&dns=...)." 78 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {4: 0, 98: 1, 5: 1, 99: 2, 64: 2, 54: 2, 12: 3, 53: 3, 107: 4, 106: 5} {78: 0, 142: 1, 65: 1, 144: 2, 61: 2, 131: 2, 20: 3, 199: 3} "The requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical JSON-based content type. The application/simpledns+json type used by this example is currently fictitious." "The discussion of Accept : header tries too hard to suggest that there will be future formats . We do n't actually know that , and the examples in the current draft have confused some developers . See PR Yep , easier to narrow this down than address the implied questions ." 19284506c3fb726148e1a62354c034830e8a12bd 19284506c3fb726148e1a62354c034830e8a12bd draft-ietf-doh-dns-over-https-latest 07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78 ['07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78', '021c92cc72fb7bc7a922307a0c01249da467139a386e780da9116a4780bae75f'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/78. A "the requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical json-based content type. the application/simpledns+json type used by this example is currently fictitious." 78 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {4: 0, 98: 1, 5: 1, 99: 2, 64: 2, 54: 2, 12: 3, 53: 3, 107: 4, 106: 5} {78: 0, 142: 1, 65: 1, 144: 2, 61: 2, 131: 2, 20: 3, 199: 3} "The requests are represented as application/dns-udpwirefomat typed bodies." "The discussion of Accept : header tries too hard to suggest that there will be future formats . We do n't actually know that , and the examples in the current draft have confused some developers . See PR Yep , easier to narrow this down than address the implied questions ." 19284506c3fb726148e1a62354c034830e8a12bd 19284506c3fb726148e1a62354c034830e8a12bd draft-ietf-doh-dns-over-https-latest 021c92cc72fb7bc7a922307a0c01249da467139a386e780da9116a4780bae75f ['07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78', '021c92cc72fb7bc7a922307a0c01249da467139a386e780da9116a4780bae75f'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/78. B "the requests are represented as application/dns-udpwirefomat typed bodies." 78 {38, 70, 19, 20, 118, 94} {19: 0, 118: 1, 94: 1, 70: 2, 38: 2, 20: 2} {78: 0, 142: 1, 43: 1, 144: 2, 122: 2, 1: 2} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that at least one JSON-based response format will be defined in the future." "The discussion of Accept : header tries too hard to suggest that there will be future formats . We do n't actually know that , and the examples in the current draft have confused some developers . See PR Yep , easier to narrow this down than address the implied questions ." 19284506c3fb726148e1a62354c034830e8a12bd 19284506c3fb726148e1a62354c034830e8a12bd draft-ietf-doh-dns-over-https-latest 1eb67f6f517e4fc148cf143742bc0bb28d91ed4cf0dc83dc4d13ee67a16d98e1 ['27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a', '1eb67f6f517e4fc148cf143742bc0bb28d91ed4cf0dc83dc4d13ee67a16d98e1'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/78. A "at the time this is published, the response types are works in progress. the only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that at least one json-based response format will be defined in the future." 78 {38, 70, 19, 20, 118, 94} {19: 0, 118: 1, 94: 1, 70: 2, 38: 2, 20: 2} {78: 0, 142: 1, 43: 1, 144: 2, 122: 2, 1: 2} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that other response formats will be defined in the future." "The discussion of Accept : header tries too hard to suggest that there will be future formats . We do n't actually know that , and the examples in the current draft have confused some developers . See PR Yep , easier to narrow this down than address the implied questions ." 19284506c3fb726148e1a62354c034830e8a12bd 19284506c3fb726148e1a62354c034830e8a12bd draft-ietf-doh-dns-over-https-latest 27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a ['27153c77f6694b6e2dc2804b76c5089128e33fcf08f8936931032312b4629e3a', '1eb67f6f517e4fc148cf143742bc0bb28d91ed4cf0dc83dc4d13ee67a16d98e1'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/78. B "at the time this is published, the response types are works in progress. the only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that other response formats will be defined in the future." 76 {80, 115} {115: 0, 80: 1} {76: 0, 146: 1} "Two primary uses cases were considered during this protocol's development. They included preventing on-path devices from interfering with DNS operations and allowing web applications to access DNS information via existing browser APIs in a safe way consistent with Cross Origin Resource Sharing (CORS) CORS. There are certainly other uses for this work." "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest f8a2280983a1e319e157524334815a3adc7354805fa7e801feb499216bf9e2c6 ['f8a2280983a1e319e157524334815a3adc7354805fa7e801feb499216bf9e2c6'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 B "two primary uses cases were considered during this protocol's development. they included preventing on-path devices from interfering with dns operations and allowing web applications to access dns information via existing browser apis in a safe way consistent with cross origin resource sharing (cors) cors. there are certainly other uses for this work." 76 {56, 83, 39} {39: 0, 56: 1, 83: 1} {76: 0, 43: 1, 41: 1} There are two initial use cases for this protocol. "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest 5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1 ['5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1', '09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9', '7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', 'bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', '0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16', 'e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 A there are two initial use cases for this protocol. 76 {56, 83, 39} {39: 0, 56: 1, 83: 1} {76: 0, 43: 1, 41: 1} "The primary use case is to prevent on-path network devices from interfering with DNS operations. This interference includes, but is not limited to, spoofing DNS responses, blocking DNS requests, and tracking." "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa ['5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1', '09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9', '7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', 'bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', '0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16', 'e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 A "the primary use case is to prevent on-path network devices from interfering with dns operations. this interference includes, but is not limited to, spoofing dns responses, blocking dns requests, and tracking." 76 {56, 83, 39} {39: 0, 56: 1, 83: 1} {76: 0, 43: 1, 41: 1} "In this use, clients - whether operating systems or individual applications - will be explicitly configured to use a DOH server as a recursive resolver by its user (or administrator). They might use the DOH server for all queries, or only for a subset of them. The specific configuration mechanism is out of scope for this document." "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest 0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16 ['5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1', '09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9', '7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', 'bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', '0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16', 'e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 A "in this use, clients - whether operating systems or individual applications - will be explicitly configured to use a doh server as a recursive resolver by its user (or administrator). they might use the doh server for all queries, or only for a subset of them. the specific configuration mechanism is out of scope for this document." 76 {56, 83, 39} {39: 0, 56: 1, 83: 1} {76: 0, 43: 1, 41: 1} "A secondary use case is allowing web applications to access DNS information, by using existing APIs in browsers to access it over HTTP in a safe way consistent with Cross Origin Resource Sharing (CORS) CORS." "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8 ['5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1', '09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9', '7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', 'bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', '0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16', 'e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 A "a secondary use case is allowing web applications to access dns information, by using existing apis in browsers to access it over http in a safe way consistent with cross origin resource sharing (cors) cors." 76 {56, 83, 39} {39: 0, 56: 1, 83: 1} {76: 0, 43: 1, 41: 1} "This is technically already possible (since the server controls both the HTTP resources it exposes and the use of browser APIs by its content), but standardization might make this easier to accomplish." "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest 7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206 ['5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1', '09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9', '7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', 'bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', '0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16', 'e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 A "this is technically already possible (since the server controls both the http resources it exposes and the use of browser apis by its content), but standardization might make this easier to accomplish." 76 {56, 83, 39} {39: 0, 56: 1, 83: 1} {76: 0, 43: 1, 41: 1} "Note that in this second use, the browser does not consult the DOH server or use its responses for any DNS lookups outside the scope of the application using them; i.e., there is (currently) no API that allows a Web site to poison DNS for others." "I think I managed to make this a tad tighter .. merge it if you agree Some people reading the document thought that it only applied to the two use cases given , and could not be used for other things . It might be good to just pare it down to its essentials and move it as a sub - section of the Introduction . See PR . closed by 76" 220449aeedf0073216d4c15a6fe83ff6974362ea 220449aeedf0073216d4c15a6fe83ff6974362ea draft-ietf-doh-dns-over-https-latest 09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9 ['5fff326dada4f0cd2dd882274420ea421791365853507ab9592ed726a396d9c1', '09e1b906f7d1c7daf24ffcf3be2a6eaf4b3a6336f3f6a720c949adbf14bffbe9', '7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', 'bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', '0a01ded4bcf4026c1d4e7d68548aab8a36a977b298b63de5afb90ba7a67e1f16', 'e51589b153427c664cd8bcc504b3ca2b6c85e3479670d1ef5325a40cf55b16fa'] https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/76 A "note that in this second use, the browser does not consult the doh server or use its responses for any dns lookups outside the scope of the application using them; i.e., there is (currently) no api that allows a web site to poison dns for others." 67 {} {} {} "The working group will analyze the security and privacy issues that could arise from accessing DNS over HTTPS. In particular, the working group will consider the interaction of DNS and HTTP caching." Also remove extraneous comment about discovery in the charter b7120ea449cccfcbc72bd52fcc1807217672f781 b7120ea449cccfcbc72bd52fcc1807217672f781 draft-ietf-doh-dns-over-https-latest 36864fab030cc7e349f3e9e5c2773f45520effc393ba9a78d8950b913b6533c0 ['99bff12554b79c8cc87af9071fc8cc704c0ad2ef49cd50771eade2e35e2135a7', '36864fab030cc7e349f3e9e5c2773f45520effc393ba9a78d8950b913b6533c0'] A "the working group will analyze the security and privacy issues that could arise from accessing dns over https. in particular, the working group will consider the interaction of dns and http caching." 67 {} {} {} "caching describes the interaction of this protocol with HTTP caching. An adversary that can control the cache used by the client can affect that client's view of the DNS. This is no different than the security implications of HTTP caching for other protocols that use HTTP." Also remove extraneous comment about discovery in the charter b7120ea449cccfcbc72bd52fcc1807217672f781 b7120ea449cccfcbc72bd52fcc1807217672f781 draft-ietf-doh-dns-over-https-latest 99bff12554b79c8cc87af9071fc8cc704c0ad2ef49cd50771eade2e35e2135a7 ['99bff12554b79c8cc87af9071fc8cc704c0ad2ef49cd50771eade2e35e2135a7', '36864fab030cc7e349f3e9e5c2773f45520effc393ba9a78d8950b913b6533c0'] B "caching describes the interaction of this protocol with http caching. an adversary that can control the cache used by the client can affect that client's view of the dns. this is no different than the security implications of http caching for other protocols that use http." 67 {} {} {} "The working group may define mechanisms for discovery of DOH servers similar to existing mechanisms for discovering other DNS servers if the chairs determine that there is both sufficient interest and working group consensus." Also remove extraneous comment about discovery in the charter b7120ea449cccfcbc72bd52fcc1807217672f781 b7120ea449cccfcbc72bd52fcc1807217672f781 draft-ietf-doh-dns-over-https-latest 3db1be71250d99a370e47f51053d502c6de7db80deebfa160a485fd7a038e42f ['3db1be71250d99a370e47f51053d502c6de7db80deebfa160a485fd7a038e42f'] A "the working group may define mechanisms for discovery of doh servers similar to existing mechanisms for discovering other dns servers if the chairs determine that there is both sufficient interest and working group consensus." 65 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {5: 0, 64: 1, 4: 1, 54: 1, 12: 2, 98: 2, 53: 2, 99: 3, 107: 3, 106: 4} {65: 0, 61: 1, 78: 1, 131: 1, 20: 2, 142: 2, 199: 2, 144: 3} "These examples use a DNS API service located at https://dnsserver.example.net/dns-query. The requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical JSON-based content type. The application/simpledns+json type used by this example is currently fictitious." "I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." f6f58337d9478877a0e5421a5cb0ac2b4de9a385 f6f58337d9478877a0e5421a5cb0ac2b4de9a385 draft-ietf-doh-dns-over-https-latest 3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26 ['07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78', '74e7d04d9add819b5e0608c1ff04f6833689b6bc9924e81bf658b9a24626dc6d', '3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26', '795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" A "these examples use a dns api service located at https://dnsserver.example.net/dns-query. the requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical json-based content type. the application/simpledns+json type used by this example is currently fictitious." 65 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {5: 0, 64: 1, 4: 1, 54: 1, 12: 2, 98: 2, 53: 2, 99: 3, 107: 3, 106: 4} {65: 0, 61: 1, 78: 1, 131: 1, 20: 2, 142: 2, 199: 2, 144: 3} "These examples use a DNS API service located at https://dnsserver.example.net/dns-query to resolve the IN A records." "I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." f6f58337d9478877a0e5421a5cb0ac2b4de9a385 f6f58337d9478877a0e5421a5cb0ac2b4de9a385 draft-ietf-doh-dns-over-https-latest 795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53 ['07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78', '74e7d04d9add819b5e0608c1ff04f6833689b6bc9924e81bf658b9a24626dc6d', '3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26', '795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" B "these examples use a dns api service located at https://dnsserver.example.net/dns-query to resolve the in a records." 65 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {5: 0, 64: 1, 4: 1, 54: 1, 12: 2, 98: 2, 53: 2, 99: 3, 107: 3, 106: 4} {65: 0, 61: 1, 78: 1, 131: 1, 20: 2, 142: 2, 199: 2, 144: 3} "The requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical JSON-based content type. The application/simpledns+json type used by this example is currently fictitious." "I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." f6f58337d9478877a0e5421a5cb0ac2b4de9a385 f6f58337d9478877a0e5421a5cb0ac2b4de9a385 draft-ietf-doh-dns-over-https-latest 07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78 ['07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78', '74e7d04d9add819b5e0608c1ff04f6833689b6bc9924e81bf658b9a24626dc6d', '3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26', '795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" B "the requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical json-based content type. the application/simpledns+json type used by this example is currently fictitious." 65 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {5: 0, 64: 1, 4: 1, 54: 1, 12: 2, 98: 2, 53: 2, 99: 3, 107: 3, 106: 4} {65: 0, 61: 1, 78: 1, 131: 1, 20: 2, 142: 2, 199: 2, 144: 3} "Finally, a GET based query for a.62characterlabel-makes-base64url- distinct-from-standard-base64.example.com is shown as an example to emphasize that the encoding alphabet of base64url is different than regular base64 and that padding is omitted." "I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." f6f58337d9478877a0e5421a5cb0ac2b4de9a385 f6f58337d9478877a0e5421a5cb0ac2b4de9a385 draft-ietf-doh-dns-over-https-latest 74e7d04d9add819b5e0608c1ff04f6833689b6bc9924e81bf658b9a24626dc6d ['07fdd0714091c38fbb8a3a0e72b1ffcd8851543a397ddade32183544d73b2d78', '74e7d04d9add819b5e0608c1ff04f6833689b6bc9924e81bf658b9a24626dc6d', '3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26', '795597d6c6f4e6c9d9203bfe7d48f962e5fd264bbd833305f5153b3b97636c53'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" B "finally, a get based query for a.62characterlabel-makes-base64url- distinct-from-standard-base64.example.com is shown as an example to emphasize that the encoding alphabet of base64url is different than regular base64 and that padding is omitted." 64 {96, 22, 23, 58, 91} {96: 0, 91: 1, 22: 1, 23: 1, 58: 2} {64: 0, 80: 1, 60: 1, 43: 1, 19: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the body parameter. The value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. An empty value indicates the default application/dns- udpwireformat type (e.g. ct&body=...)." "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e ['cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '09203511966bf6b28399001c6d6f7de2a81d25d6bc60e04b5b95429078cbd484', 'e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" A "when using the get method the uri path must contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the body parameter. the value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. an empty value indicates the default application/dns- udpwireformat type (e.g. ct&body=...)." 64 {96, 22, 23, 58, 91} {96: 0, 91: 1, 22: 1, 23: 1, 58: 2} {64: 0, 80: 1, 60: 1, 43: 1, 19: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""body"". The value of the parameter is the content of the request potentially encoded with base64url RFC4648. Specifications that define media types for use with DOH, such as DNS Wire Format dnswire of this document, MUST indicate if the body parameter uses base64url encoding." "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest 6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a ['cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '09203511966bf6b28399001c6d6f7de2a81d25d6bc60e04b5b95429078cbd484', 'e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" A "when using the get method the uri path must contain a query parameter with the name of ""body"". the value of the parameter is the content of the request potentially encoded with base64url rfc4648. specifications that define media types for use with doh, such as dns wire format dnswire of this document, must indicate if the body parameter uses base64url encoding." 64 {96, 22, 23, 58, 91} {96: 0, 91: 1, 22: 1, 23: 1, 58: 2} {64: 0, 80: 1, 60: 1, 43: 1, 19: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the dns parameter. The value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&dns=...) or it may be empty. An empty value indicates the default application/dns- udpwireformat type (e.g. ct&dns=...)." "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d ['cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '09203511966bf6b28399001c6d6f7de2a81d25d6bc60e04b5b95429078cbd484', 'e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" B "when using the get method the uri path must contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the dns parameter. the value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&dns=...) or it may be empty. an empty value indicates the default application/dns- udpwireformat type (e.g. ct&dns=...)." 64 {96, 22, 23, 58, 91} {96: 0, 91: 1, 22: 1, 23: 1, 58: 2} {64: 0, 80: 1, 60: 1, 43: 1, 19: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""dns"". The value of the parameter is the content of the request potentially encoded with base64url RFC4648. Specifications that define media types for use with DOH, such as DNS Wire Format dnswire of this document, MUST indicate if the body parameter uses base64url encoding." "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest 09203511966bf6b28399001c6d6f7de2a81d25d6bc60e04b5b95429078cbd484 ['cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '09203511966bf6b28399001c6d6f7de2a81d25d6bc60e04b5b95429078cbd484', 'e541517e7b9a0a69115903000fb96b3f9e090e88b3656f7f011fff4e75d2095d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" B "when using the get method the uri path must contain a query parameter with the name of ""dns"". the value of the parameter is the content of the request potentially encoded with base64url rfc4648. specifications that define media types for use with doh, such as dns wire format dnswire of this document, must indicate if the body parameter uses base64url encoding." 64 {} {} {} "When using the GET method, the body MUST be encoded with base64url RFC4648." "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest 09108b5e90091f18dc298d139ff97d4b69bc89aaea108c808a67acb8ee6fa804 ['09108b5e90091f18dc298d139ff97d4b69bc89aaea108c808a67acb8ee6fa804', '041c931fb0d6c49c7fe77ded32ab07936e9855c79401c5bed967efb1f3baec33', '902f600775128a77a54362510e7c222a638f1b4a5b1b0357c9e7525a7a80dee4'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" A "when using the get method, the body must be encoded with base64url rfc4648." 64 {} {} {} Padding characters for base64url MUST NOT be included. "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest 902f600775128a77a54362510e7c222a638f1b4a5b1b0357c9e7525a7a80dee4 ['09108b5e90091f18dc298d139ff97d4b69bc89aaea108c808a67acb8ee6fa804', '041c931fb0d6c49c7fe77ded32ab07936e9855c79401c5bed967efb1f3baec33', '902f600775128a77a54362510e7c222a638f1b4a5b1b0357c9e7525a7a80dee4'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" A padding characters for base64url must not be included. 64 {} {} {} "When using the GET method, the body MUST be encoded with base64url RFC4648 and then placed as a name value pair in the query portion of the URI with name ""dns"". Padding characters for base64url MUST NOT be included." "partially addresses is right - the naming of this query parameter was just asking for pain . new coat of paint is in order . : +1 : I 've seen an implementation report that confused and base64url ; they noted that the examples did n't include one where the distinction was important yes , there is also the bit about not having the sign paddings that caught me at first . During the DOH proxy implementation , I missed the padding part . I have it implemented in and I think it is correct :) Would an example implementation in a language be worth adding to the examples ? It may be more straightforward to read than examples . documentation says : This may be worth highlighting as what base64url is too . This is complicated by the fact that some people want the encoding of responses to GET to be media - type - specific . I think you mean GET request body parameters ( not responses - which are always in bodies and not encoded ) Yes , I meant body parameters . Please stop calling the parameter . It 's really confusing . ( I just read this and thought to myself "" GET does n't have a body "" ) . note that the current encoded url in the github copy is wrong as it was n't updated when the dns i d changed to 0 . I 'll fix that as part of making a base 64 example ." c9b799086b7d231c128564cb4308953f3142c586 c9b799086b7d231c128564cb4308953f3142c586 draft-ietf-doh-dns-over-https-latest 041c931fb0d6c49c7fe77ded32ab07936e9855c79401c5bed967efb1f3baec33 ['09108b5e90091f18dc298d139ff97d4b69bc89aaea108c808a67acb8ee6fa804', '041c931fb0d6c49c7fe77ded32ab07936e9855c79401c5bed967efb1f3baec33', '902f600775128a77a54362510e7c222a638f1b4a5b1b0357c9e7525a7a80dee4'] "https://github.com/chantra/doh-proxy/blob/master/dohproxy/utils.py https://docs.python.org/3/library/base64.html#base64.urlsafe_b64encode" B "when using the get method, the body must be encoded with base64url rfc4648 and then placed as a name value pair in the query portion of the uri with name ""dns"". padding characters for base64url must not be included." 63 {32, 101, 102, 103, 40, 44, 45, 31} {31: 0, 32: 1, 40: 2, 103: 3, 45: 4, 102: 5, 101: 6, 44: 7} {63: 0, 157: 1, 174: 2, 199: 3, 169: 4, 144: 5, 142: 6, 128: 7} "Before using DOH response data for DNS resolution, the client MUST establish that the HTTP request URI is a trusted service for the DOH query. For HTTP requests initiated by the DNS API client this trust is implicit in the selection of URI. For HTTP server push (RFC7540 Section 8.2) extra care must be taken to ensure that the pushed URI is one that the client would have directed the same query to if the client had initiated the request. This specification does not extend DNS resolution privileges to URIs that are not recognized by the client as trusted DNS API servers." "Describe how GET ( not POST ) is used with server push . This will ensure that you answer the bigger question : what do you do with a server push from an arbitrary server ? I assume that this wo n't be privileged in any way , and only the configured DNS API server will have its answers routed to the DNS stack , but that needs to be explicit . Server push in a request - response protocol like DNS would take a lot of designing . It is n't really needed in DNS because the server can always slap in additional answers in the Additional section of a request . I propose we close this by saying "" HTTP server push is not defined for this protocol , but might be defined in a future version "" . If someone wants to define an EDNS0 or session - signal extension for it , that should be just fine . HTTP already has server push , so I would prefer something closer to "" this protocol does not alter the semantics of HTTP server push "" . I think is closer to what I 'm thinking .. ( though I do n't really think that needs to be stated beyond other clarifications that this document just uses http in a particular manner , it does not define it . ) additionally RR 's do n't quite address the same use case by the way as push can operate in more of a pub - sub fashion if you like where an additional RR is definitely a poll model . Of course this protocol does not alter the semantics of server push , but what does it mean for a DNS client ? In the DNS protocol , the only thing even vaguely like server - initiated information is stuff in the Additional section of an answer , and even then , there are pretty strict rules about ignoring it if it is at all smelly . an often times misunderstood part of push is that a server pushes both a response and the request that it is responding to . ( like dns , http is always a request / response pair , so if the client does n't provide the request the server needs to ) . So that 's what it means - its an answer to a particular question and you can cache it or not depending on whether that 's a request you would have asked / trusted the peer to handle . That 's pretty much the same rule for dns and traditional http clients . A client who would rather not have things pushed at them has existing protocol mechanisms to disable that ( SETTINGS ) which doh does n't need to specify . I think all of this gets at what was originally commenting on - you need to make a decision on whether or not you trust the server for the query / response being pushed ( i.e is this your dns api server otherwise do n't accept ) before processing it .. this is already true for anything pushed via http but a sentence making it explicit is fine ." b0bf256a13d965962b85ac4b34ee253170ab7a95 b0bf256a13d965962b85ac4b34ee253170ab7a95 draft-ietf-doh-dns-over-https-latest 48cd3653ab75fd1e0d9d7cc8d08a5c555f4209a8404b6d4ff99167d9650e387e ['48cd3653ab75fd1e0d9d7cc8d08a5c555f4209a8404b6d4ff99167d9650e387e'] B "before using doh response data for dns resolution, the client must establish that the http request uri is a trusted service for the doh query. for http requests initiated by the dns api client this trust is implicit in the selection of uri. for http server push (rfc7540 section 8.2) extra care must be taken to ensure that the pushed uri is one that the client would have directed the same query to if the client had initiated the request. this specification does not extend dns resolution privileges to uris that are not recognized by the client as trusted dns api servers." 62 {6} {6: 0} {62: 0} "A DNS API client may face a similar bootstrapping problem when the HTTP request needs to resolve the hostname portion of the DNS URI. Just as the address of a traditional DNS nameserver cannot be originally determined from that same server, a DOH client cannot use its DOH server to initially resolve the server's host name into an address. Alternative strategies a client might employ include making the initial resolution part of the configuration, IP based URIs and corresponding IP based certificates for HTTPS, or resolving the DNS API Server's hostname via traditional DNS or another DOH server while still authenticating the resulting connection via HTTPS." "this is a little longer than I had hoped given that bootstrap deadlock is not a new problem for dns introduced by doh .. but the is worth mentioning . If DOH uses https under the hood , how does this https host name get resolved ? I would have expected this to be mentioned in the RFC , though I did n't see any mention of it , besides OCSP Stapling . Is it assumed that DOH would need to be bootstrapped by traditional DNS ? Or is it assumed that a DOH endpoint would have a TLS / SSL certificate for its ip address ? ( Apparently a certificate for an IP address is possible in some cases , though hard to get . ) I assume DOH is not intended to be used by home NAT routers ? ( How would the router get a signed TLS certificate ? Maybe a DHCP response could say : use this DOH url and trust this self - signed cert for its ip address ( though a certificate might be too big for a DHCP response ) . Or , in cases where no local DNS services are needed , the DHCP response could maybe say : use this public DOH url , here 's the ip address for that hostname ) The document does n't describe how a HTTPS client authenticates HTTPS ... that 's on purpose in order to simply leverage HTTPS without surprises . as a practical matter a number of things can be done . One is TOFU as you suggest . Another is an IP based certificate which you also suggest . Another would be to lookup the name via another , perhaps less trusted resolver , knowing that HTTPS means the IP address is n't part of the security scheme anyhow - its just a route .. subject to blocking , but not to impersonation . Another would be to bootstrap the client with a name and an address instead of one or the other . but as a general matter I think this is one step beyond the scope of the document as its basically writing down the rules for HTTPS . This issue semi - duplicates I think this should at least be called out as an operational concern . e.g. mentioning that you probably need an alternate resolver to find the DOH server , after which a paranoid implementation can re - query for the DOH 's address as an integrity check . This is the direction I have taken for . When opening a connection to the remote DOH server , if an ip was provided , it will just use that , if not , it will try to resolve using the local stub , but that may not work .... Just like an IP has always been provided to bootstrap which nameserver to use , here we need to also be provided a name so we can validate that the server we are talking to is the one we trust ." 98d0381ede2e53d06840a67da69e6d9a6c3ae835 98d0381ede2e53d06840a67da69e6d9a6c3ae835 draft-ietf-doh-dns-over-https-latest 08923af3ae7a208da9326d0d1ddd71204c6c8ce5d733262a1fda9215d63adc1d ['08923af3ae7a208da9326d0d1ddd71204c6c8ce5d733262a1fda9215d63adc1d'] B "a dns api client may face a similar bootstrapping problem when the http request needs to resolve the hostname portion of the dns uri. just as the address of a traditional dns nameserver cannot be originally determined from that same server, a doh client cannot use its doh server to initially resolve the server's host name into an address. alternative strategies a client might employ include making the initial resolution part of the configuration, ip based uris and corresponding ip based certificates for https, or resolving the dns api server's hostname via traditional dns or another doh server while still authenticating the resulting connection via https." 61 {65, 67, 36, 73, 10, 75, 113} {65: 0, 73: 1, 10: 1, 75: 1, 67: 2, 36: 2, 113: 2} {61: 0, 131: 1, 6: 1, 33: 1, 133: 2, 128: 2, 199: 2} "To make a DNS API query, a DNS API client sends an HTTP request to the URI of the DNS API." "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b ['8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', '6196e3f05369821215cda9627a6a8ad4b3aaae6b2722afc4e55d2bdb87455df6', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d'] A "to make a dns api query, a dns api client sends an http request to the uri of the dns api." 61 {65, 67, 36, 73, 10, 75, 113} {65: 0, 73: 1, 10: 1, 75: 1, 67: 2, 36: 2, 113: 2} {61: 0, 131: 1, 6: 1, 33: 1, 133: 2, 128: 2, 199: 2} The URI scheme MUST be https. "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest 6196e3f05369821215cda9627a6a8ad4b3aaae6b2722afc4e55d2bdb87455df6 ['8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', '6196e3f05369821215cda9627a6a8ad4b3aaae6b2722afc4e55d2bdb87455df6', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d'] A the uri scheme must be https. 61 {65, 67, 36, 73, 10, 75, 113} {65: 0, 73: 1, 10: 1, 75: 1, 67: 2, 36: 2, 113: 2} {61: 0, 131: 1, 6: 1, 33: 1, 133: 2, 128: 2, 199: 2} "A client can be configured with a DNS API URI, or it can discover the URI. This document defines a well-known URI path of ""/.well-known/ dns-query"" so that a discovery process that produces a domain name or domain name and port can be used to construct the DNS API URI. (See iana for the registration of this in the well-known URI registry.) DNS API servers SHOULD use this well-known path to help contextualize DNS Query requests that use server push RFC7540." "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest 10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d ['8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', '6196e3f05369821215cda9627a6a8ad4b3aaae6b2722afc4e55d2bdb87455df6', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d'] A "a client can be configured with a dns api uri, or it can discover the uri. this document defines a well-known uri path of ""/.well-known/ dns-query"" so that a discovery process that produces a domain name or domain name and port can be used to construct the dns api uri. (see iana for the registration of this in the well-known uri registry.) dns api servers should use this well-known path to help contextualize dns query requests that use server push rfc7540." 61 {65, 67, 36, 73, 10, 75, 113} {65: 0, 73: 1, 10: 1, 75: 1, 67: 2, 36: 2, 113: 2} {61: 0, 131: 1, 6: 1, 33: 1, 133: 2, 128: 2, 199: 2} "A DNS API Client encodes a single DNS query into the HTTP request using either the HTTP GET or POST methods." "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest 8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e ['8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', '6196e3f05369821215cda9627a6a8ad4b3aaae6b2722afc4e55d2bdb87455df6', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d'] A "a dns api client encodes a single dns query into the http request using either the http get or post methods." 61 {65, 67, 36, 73, 10, 75, 113} {65: 0, 73: 1, 10: 1, 75: 1, 67: 2, 36: 2, 113: 2} {61: 0, 131: 1, 6: 1, 33: 1, 133: 2, 128: 2, 199: 2} "To make a DNS API query a DNS API client encodes a single DNS query into an HTTP request using either the HTTP GET or POST method and the other requirements of this section. The DNS API server defines the URI used by the request. Configuration and discovery of the URI is done out of band from this protocol." "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157 ['8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', '6196e3f05369821215cda9627a6a8ad4b3aaae6b2722afc4e55d2bdb87455df6', 'c1fb7c1b15821930be54468a16d1f731e3b405730dc06973b8d4832fb6167157', '10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d'] B "to make a dns api query a dns api client encodes a single dns query into an http request using either the http get or post method and the other requirements of this section. the dns api server defines the uri used by the request. configuration and discovery of the uri is done out of band from this protocol." 61 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {64: 0, 12: 1, 5: 1, 4: 2, 54: 2, 98: 3, 53: 3, 99: 4, 107: 4, 106: 5} {61: 0, 20: 1, 65: 1, 78: 2, 131: 2, 142: 3, 199: 3, 144: 4} "For this example assume a DNS API server is following this specification on origin https://dnsserver.example.net/ and the well- known path. The DNS API client chooses to send its requests in application/dns-udpwirefomat but indicates it can parse replies in that format or as a hypothetical JSON-based content type. The application/simpledns+json type used by this example is currently fictitious." "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest 8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2 ['8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2', '3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26'] A "for this example assume a dns api server is following this specification on origin https://dnsserver.example.net/ and the well- known path. the dns api client chooses to send its requests in application/dns-udpwirefomat but indicates it can parse replies in that format or as a hypothetical json-based content type. the application/simpledns+json type used by this example is currently fictitious." 61 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {64: 0, 12: 1, 5: 1, 4: 2, 54: 2, 98: 3, 53: 3, 99: 4, 107: 4, 106: 5} {61: 0, 20: 1, 65: 1, 78: 2, 131: 2, 142: 3, 199: 3, 144: 4} "These examples use a DNS API service located at https://dnsserver.example.net/dns-query. The requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical JSON-based content type. The application/simpledns+json type used by this example is currently fictitious." "and The draft says : But it probably needs to describe at least one way to perform that automated discovery . It 's probably not a good idea to have people attempt to guess . fwiw the .wk is used here so that that when a request / response pair is pushed the client understands this to be DoH. Its probably also helpful to have a well - known path here for the normal so that defaults of https:// , 443 , and the default path can be used in configuration systems that might only be capable of a hostname . ( e.g. my DoH server is at ) .. I 'll work on text . I think I can future proof for the context I want using a response header instead of a well known .. imo that 's kind of ignoring the point of the path - but whatever . Maybe I 'll propose x - well - known :)" 15826c57f9ec6a1902772acf94ceb9690b95645f 15826c57f9ec6a1902772acf94ceb9690b95645f draft-ietf-doh-dns-over-https-latest 3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26 ['8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2', '3172d3ee00b0d4d18f7df71de34f0d4c7a348c07a164ca38691cb84d58912c26'] B "these examples use a dns api service located at https://dnsserver.example.net/dns-query. the requests are represented as application/dns-udpwirefomat typed bodies, but the client indicates it can parse responses in either that format or as a hypothetical json-based content type. the application/simpledns+json type used by this example is currently fictitious." 60 {96, 22, 23, 58, 91} {22: 0, 96: 1, 23: 1, 91: 2, 58: 2} {60: 0, 64: 1, 43: 1, 80: 2, 19: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""body"". The value of the parameter is the content of the request encoded with base64url RFC4648. Using the GET method is friendlier to many HTTP cache implementations." "closes issue this is an evolution of" ba41be5465b98ecfbdea46c13f9366e1b97534b7 ba41be5465b98ecfbdea46c13f9366e1b97534b7 draft-ietf-doh-dns-over-https-latest 339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4 ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', '1950961aeb77e50a1017321599e53ec5b4e6fdaec63570f7fff9b3214cfe4a2d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] A "when using the get method the uri path must contain a query parameter with the name of ""body"". the value of the parameter is the content of the request encoded with base64url rfc4648. using the get method is friendlier to many http cache implementations." 60 {96, 22, 23, 58, 91} {22: 0, 96: 1, 23: 1, 91: 2, 58: 2} {60: 0, 64: 1, 43: 1, 80: 2, 19: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""body"". The value of the parameter is the content of the request potentially encoded with base64url RFC4648. Specifications that define media types for use with DOH, such as DNS Wire Format dnswire of this document, MUST indicate if the body parameter uses base64url encoding." "closes issue this is an evolution of" ba41be5465b98ecfbdea46c13f9366e1b97534b7 ba41be5465b98ecfbdea46c13f9366e1b97534b7 draft-ietf-doh-dns-over-https-latest 6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', '1950961aeb77e50a1017321599e53ec5b4e6fdaec63570f7fff9b3214cfe4a2d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] B "when using the get method the uri path must contain a query parameter with the name of ""body"". the value of the parameter is the content of the request potentially encoded with base64url rfc4648. specifications that define media types for use with doh, such as dns wire format dnswire of this document, must indicate if the body parameter uses base64url encoding." 60 {96, 22, 23, 58, 91} {22: 0, 96: 1, 23: 1, 91: 2, 58: 2} {60: 0, 64: 1, 43: 1, 80: 2, 19: 2} "Using the GET method is friendlier to many HTTP cache implementations." "closes issue this is an evolution of" ba41be5465b98ecfbdea46c13f9366e1b97534b7 ba41be5465b98ecfbdea46c13f9366e1b97534b7 draft-ietf-doh-dns-over-https-latest 1950961aeb77e50a1017321599e53ec5b4e6fdaec63570f7fff9b3214cfe4a2d ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', '1950961aeb77e50a1017321599e53ec5b4e6fdaec63570f7fff9b3214cfe4a2d', '6fbe84bf49d7948be74b5e8fc3c58d713f3f93a76ebd32aaafb572c026c86d9a'] B "using the get method is friendlier to many http cache implementations." 56 {33, 34, 18, 120, 26, 30} {18: 0, 26: 1, 30: 1, 33: 2, 34: 3, 120: 4} {56: 0, 39: 1, 101: 1, 33: 2, 21: 3, 42: 4} "The Answer section of a DNS response contains one or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset is supposed to have the Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs, though it is only possible for the HTTP response to have a single freshness lifetime. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the Resource Record bearing the smallest TTL in the Answer section of the response. The HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest 25bbe1793ffe189c147ca76054675bf702f8cb9e8889d95ebc5b47e5e14d1fdd ['25bbe1793ffe189c147ca76054675bf702f8cb9e8889d95ebc5b47e5e14d1fdd', '8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw A "the answer section of a dns response contains one or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset is supposed to have the time to live (ttl) freshness information. different rrsets in the answer section can have different ttls, though it is only possible for the http response to have a single freshness lifetime. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the resource record bearing the smallest ttl in the answer section of the response. the http freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {33, 34, 18, 120, 26, 30} {18: 0, 26: 1, 30: 1, 33: 2, 34: 3, 120: 4} {56: 0, 39: 1, 101: 1, 33: 2, 21: 3, 42: 4} "The Answer section of a DNS response contains one or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset is supposed to have the Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs, though it is only possible for the HTTP response to have a single freshness lifetime. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the Resource Record bearing the smallest TTL in the Answer section of the response. Specifically, the HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest 8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0 ['25bbe1793ffe189c147ca76054675bf702f8cb9e8889d95ebc5b47e5e14d1fdd', '8fdc40f6da7a73747a84fcc4db3db0ba2f49e2467fd16fe4bacfde46857bdba0'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B "the answer section of a dns response contains one or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset is supposed to have the time to live (ttl) freshness information. different rrsets in the answer section can have different ttls, though it is only possible for the http response to have a single freshness lifetime. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the resource record bearing the smallest ttl in the answer section of the response. specifically, the http freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {123} {123: 0} {56: 0} "HTTP revalidation of cached DNS information may be of limited value as revalidation provides only a bandwidth benefit and DNS transactions are normally latency bound instead. Furthermore, the HTTP response headers that enable revalidation (such as ""Last- Modified"" and ""Etag"") are often fairly large when compared to the overall DNS response size, and have a variable nature that creates constant pressure on the HTTP/2 compression dictionary RFC7541. Other types of DNS data, such as zone transfers, may be larger and benefit more from revalidation. DNS API servers may wish to consider whether providing these optional response headers is worthwhile." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest fe8aedc5f6a1e18856a051a1333f36f564ad3111c214ffd53c62b62beb2af04f ['fe8aedc5f6a1e18856a051a1333f36f564ad3111c214ffd53c62b62beb2af04f'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw A "http revalidation of cached dns information may be of limited value as revalidation provides only a bandwidth benefit and dns transactions are normally latency bound instead. furthermore, the http response headers that enable revalidation (such as ""last- modified"" and ""etag"") are often fairly large when compared to the overall dns response size, and have a variable nature that creates constant pressure on the http/2 compression dictionary rfc7541. other types of dns data, such as zone transfers, may be larger and benefit more from revalidation. dns api servers may wish to consider whether providing these optional response headers is worthwhile." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} This protocol MUST be used with https scheme URI RFC7230. "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76 ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw A this protocol must be used with https scheme uri rfc7230. Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} This protocol MUST be used with the https scheme URI RFC7230. "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598 ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B this protocol must be used with the https scheme uri rfc7230. Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} "A DOH API Client may utilize a hierarchy of caches that include both HTTP and DNS specific caches. HTTP cache entries may be bypassed with HTTP mechanisms such as the Cache-Control no-cache directive however DNS caches do not have a similar mechanism." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest 0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B "a doh api client may utilize a hierarchy of caches that include both http and dns specific caches. http cache entries may be bypassed with http mechanisms such as the cache-control no-cache directive however dns caches do not have a similar mechanism." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} "A DOH response that was previously stored in an HTTP cache will contain the RFC7234 Age response header indicating the elapsed time between when the entry was placed in the HTTP cache and the current DOH response. DNS API clients should subtract this time from the DNS TTL if they are re-sharing the information in a non HTTP context (e.g. their own DNS cache) to determine the remaining time to live of the DNS record." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B "a doh response that was previously stored in an http cache will contain the rfc7234 age response header indicating the elapsed time between when the entry was placed in the http cache and the current doh response. dns api clients should subtract this time from the dns ttl if they are re-sharing the information in a non http context (e.g. their own dns cache) to determine the remaining time to live of the dns record." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} "HTTP revalidation (e.g. via If-None-Match request headers) of cached DNS information may be of limited value to DOH as revalidation provides only a bandwidth benefit and DNS transactions are normally latency bound. Furthermore, the HTTP response headers that enable revalidation (such as ""Last-Modified"" and ""Etag"") are often fairly large when compared to the overall DNS response size, and have a variable nature that creates constant pressure on the HTTP/2 compression dictionary RFC7541. Other types of DNS data, such as zone transfers, may be larger and benefit more from revalidation. DNS API servers may wish to consider whether providing these validation enabling response headers is worthwhile." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest 677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6 ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B "http revalidation (e.g. via if-none-match request headers) of cached dns information may be of limited value to doh as revalidation provides only a bandwidth benefit and dns transactions are normally latency bound. furthermore, the http response headers that enable revalidation (such as ""last-modified"" and ""etag"") are often fairly large when compared to the overall dns response size, and have a variable nature that creates constant pressure on the http/2 compression dictionary rfc7541. other types of dns data, such as zone transfers, may be larger and benefit more from revalidation. dns api servers may wish to consider whether providing these validation enabling response headers is worthwhile." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} "The stale-while-revalidate and stale-if-error cache control directives may be well suited to a DOH implementation when allowed by server policy. Those mechanisms allow a client, at the server's discretion, to reuse a cache entry that is no longer fresh under some extenuating circumstances defined in RFC5861." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest 815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303 ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B "the stale-while-revalidate and stale-if-error cache control directives may be well suited to a doh implementation when allowed by server policy. those mechanisms allow a client, at the server's discretion, to reuse a cache entry that is no longer fresh under some extenuating circumstances defined in rfc5861." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 56 {0, 1, 74, 43, 62} {43: 0, 1: 1, 0: 2, 74: 2, 62: 2} {56: 0, 177: 1, 199: 2, 154: 2} "All HTTP servers, including DNS API servers, need to consider cache interaction when they generate responses that are not globally valid. For instance, if a DNS API server customized a response based on the client's identity then it would not want to globally allow reuse of that response. This could be accomplished through a variety of HTTP techniques such as a Cache-Control max-age of 0, or perhaps by the Vary response header." "There are plenty of reasons to "" extend "" a TTL , such as if the TTL is below a configured threshold or if one can not reach the authoritative server again . The rules for TTLs is ( again ... ) being actively debated in the DNSOP WG , and it would be inappropriate for DOH to be more restrictive than DNS itself , thus the "" SHOULD "" . Leaving a bare "" SHOULD "" there asks more questions than this text answers . If the considerations for extension are equivalent to those that apply to DNS , then saying that would help a lot . I hear that , but I 'm at a loss of what words to use , given that the words do n't exist in the DNS specs either . What is important to me is that we do n't do more than the DNS specs , as vague as they are . If implementations routinely ( or otherwise ) ignore the TTL , why not then say MUST ? Then it 's clear what the TTL is , and now they are just ignoring the TTL ( as per normal DNS practice ) . Good point , will fix before pulling . can you give a review here ? apparently wo n't let me ask you directly because you created the PR ( which I seriously amended ) The draft does n't really tackle the question of caching fully right now . I think that there is an implied hierarchy of caches that are in use . The query first hits the DNS stack , which maintains its own cache . Using the DNS TTL , that cache might decide to answer the query . Then there is an HTTP stack , which might maintain its own cache . Notably , none of the HTTP cache - busting techniques work from the perspective of the DNS client , so we have to be careful to observe that . This particularly ordering makes much worse . Finally , a DNS API server might maintain its own cache . worth noting that the DNS client does n't have e2e cache busters with an http cache in the chain .. but does a traditional dns client have a cache buster ? I 've only really seen them explicitly ask another resolver .. which still seems possible here . Could we add a sentence saying that the gateway decapsulating DNS - over - HTTPS MUST subtract the age of the HTTP response from the TTL inside the UDP packet before forwarding or consuming it ( assuming that would n't break DNSSEC ) ? I think it would help with some of the confusion about interaction between HTTP and DNS caches . probably worth noting HTTP "" Age "" and things like stale - while - revalidate to help the reader , but we want to be careful not to look like we 're making changes to http when we are n't . There was also a clear call to have full caching examples that illustrate the interactions at hand . I think my read of the room in Singapore was that we do n't want to encourage rewriting the DNS records . ( DNSSEC being at least one reason ) . I think we do want a DOH client to rewrite TTL when it is posing as a resolver to TTL - sensitive DNS clients . TTL is excluded from DNSSEC signing , so there should n't be a problem with that . on caching .... we should also clarify behaviours of caching proxies . If a caching proxy starts caching based on GET parameters , we can hit an issue when the DOH backends are providing geo - targeted based on querier IP . The first query would warm the cache and the second query from a second totally unrelated subnet would get a cached answer which is not the one it would get if it were to query the DOH server directly . It may be worth having a section about potential caching caveats . see what are you thinking of saying that is not covered by the http rfc 's ? ( 723x ) . The current document talks about how using an ID of 0 is cache friendly . Which would lead to believe that if you have a layer of caching proxies in front of the DOH servers , they could cache the content and distribute it without re - querying the DOH backend for TTL time . While this will work just fine in most cases , there is a caveat when the DOH servers return queries that are different / customized for different subnets such as with a DNS global load balancer . I am not sure if this is covered in some aspect in one of the 723X RFCs , but I think we should drop a “ consideration ” for people to realize when to disable caching from reverse proxies . If there is already an RFC detailing this , great let ’s link to it . If not I do n’t think it deserves a lot , but we should let people be aware of the potential issue . this issue is kind of broad .. I suggest we call the base fix here , and we 'll open follow on issues as necessary . Why use SHOULD ? What would cause someone to extend the TTL beyond the freshness lifetime ?" 91fe335eff9d763974267f49532fbcc334f0e41f 91fe335eff9d763974267f49532fbcc334f0e41f draft-ietf-doh-dns-over-https-latest 4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147 ['677b9715fc2700f629cb8dd84eb49304db37f2fdeafa334b154762ab7e33c1e6', 'a03519dadf00a6517aea7a290836ec735b302062778a0a1aec97bed6d6258598', '0215bf3d110cfa9cf8d71561282b4d4e6e63ef3a1ccaf094a68c5cc05bdb358a', 'a20419c676e013447f6bd74dbce1d33e1990e27b08d830e6a9c8d6fc73209eec', '4310b9601e2e927b3df334c9aad110d313d59f738211979e1111c908e7790147', 'a6ab831ec9c877588028f15dd7c3592fcba73fcfabeee10c8836f22891e89f76', '815bf040c31cd1f50444d8b1b1e94b0ebdc0566d0c6be95f1ee72b442fdd2303'] https://mailarchive.ietf.org/arch/msg/doh/DAeqQpiy5vUY_4VKms4GLIAtpuw B "all http servers, including dns api servers, need to consider cache interaction when they generate responses that are not globally valid. for instance, if a dns api server customized a response based on the client's identity then it would not want to globally allow reuse of that response. this could be accomplished through a variety of http techniques such as a cache-control max-age of 0, or perhaps by the vary response header." Hi list , Should we clarify the behaviour of caching proxies ? When using a proxy in front of DOH servers and doing targeting based on resolver IP or using ECS , the returned answer for a same query ( body param ) may be different depending on the client , but if the caching proxy caches based on GET parameters , there is chances to return answers that are supposed to be correct for another client . In draft - ietf - doh - dns - over - https-02 Section 5 : Using the GET method is friendlier to many HTTP cache implementations . and In order to maximize cache friendliness , DNS API clients using media formats that include DNS ID , such as application / dns - udpwireformat , SHOULD use a DNS ID of 0 in every DNS request . As much as we may instruct the client to cache the result , if there is proxies in the middle , those should not cache the answer they received from backend DOH servers . 50 {57, 108, 29} {57: 0, 108: 1, 29: 2} {50: 0, 106: 1, 122: 2} "The protocol must ensure interoperable media formats through a mandatory to implement format wherein a query must be able to contain one or more EDNS extensions, including those not yet defined." "Right now the text only defines that future edns options must be supported , but ( IMHO ) we should support all future versions of changes to the DNS in general , not just future EDNS extensions . So I proposed text that rewords a bit to state this and uses EDNS as an example . Good call . There might be other on - the - wire extension mechanisms in the future ." 63787142b317fa58e7289f6260908301f11a108d 63787142b317fa58e7289f6260908301f11a108d draft-ietf-doh-dns-over-https-latest 7f9b43e5625ef93e547be4848a249cd31975a4489dcb73fc2a4d2e4270af2642 ['7f9b43e5625ef93e547be4848a249cd31975a4489dcb73fc2a4d2e4270af2642', 'dcf53581e74ba5e34951581818a0244659b089eef434ba87c5f33af317ac9cdb'] A "the protocol must ensure interoperable media formats through a mandatory to implement format wherein a query must be able to contain one or more edns extensions, including those not yet defined." 50 {57, 108, 29} {57: 0, 108: 1, 29: 2} {50: 0, 106: 1, 122: 2} "The protocol must ensure interoperable media formats through a mandatory to implement format wherein a query must be able to contain future modifications to the DNS protocol including the inclusion of one or more EDNS extensions (including those not yet defined)." "Right now the text only defines that future edns options must be supported , but ( IMHO ) we should support all future versions of changes to the DNS in general , not just future EDNS extensions . So I proposed text that rewords a bit to state this and uses EDNS as an example . Good call . There might be other on - the - wire extension mechanisms in the future ." 63787142b317fa58e7289f6260908301f11a108d 63787142b317fa58e7289f6260908301f11a108d draft-ietf-doh-dns-over-https-latest dcf53581e74ba5e34951581818a0244659b089eef434ba87c5f33af317ac9cdb ['7f9b43e5625ef93e547be4848a249cd31975a4489dcb73fc2a4d2e4270af2642', 'dcf53581e74ba5e34951581818a0244659b089eef434ba87c5f33af317ac9cdb'] B "the protocol must ensure interoperable media formats through a mandatory to implement format wherein a query must be able to contain future modifications to the dns protocol including the inclusion of one or more edns extensions (including those not yet defined)." 47 {81, 66} {66: 0, 81: 1} {47: 0, 1: 1} When using the POST method, the body is not encoded. ( IMHO ) f1bfea0c915d70b500b9f47b401affdb47a9e92d f1bfea0c915d70b500b9f47b401affdb47a9e92d draft-ietf-doh-dns-over-https-latest 8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43 ['8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43', 'c3c9b26024af4a2be985bae8701e9b74b25f4ce8fe4bbe20c6d7949bf98a1a30'] A when using the post method, the body is not encoded. 47 {81, 66} {66: 0, 81: 1} {47: 0, 1: 1} When using the POST method, the body MUST NOT be encoded. ( IMHO ) f1bfea0c915d70b500b9f47b401affdb47a9e92d f1bfea0c915d70b500b9f47b401affdb47a9e92d draft-ietf-doh-dns-over-https-latest c3c9b26024af4a2be985bae8701e9b74b25f4ce8fe4bbe20c6d7949bf98a1a30 ['8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43', 'c3c9b26024af4a2be985bae8701e9b74b25f4ce8fe4bbe20c6d7949bf98a1a30'] B when using the post method, the body must not be encoded. 45 {121, 59, 61, 47} {47: 0, 59: 1, 121: 2, 61: 2} {45: 0, 34: 1, 165: 2, 1: 2} "The messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, parallelism, priority, and header compression to acheive similar performance. Those features were introduced to HTTP in HTTP/2 RFC7540. Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but would result in very poor performance for many uses cases." "This text is rather subjective without data . Changing it to "" may "" still brings the point without needing a reference to a study proving the sentence true . ( whoops ; just realized I 've been blowing my pull requests by chaining them off a single branch ... my bad , and this will cause you some pain )" 90ffdb90666c2f5fefb94f3e2bad9303a98173a8 90ffdb90666c2f5fefb94f3e2bad9303a98173a8 draft-ietf-doh-dns-over-https-latest 9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439 ['70f4ea02178a1414d64a892cfeb64e15a2cfe2f82f412e43f153c229db2d2767', '9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439'] A "the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, parallelism, priority, and header compression to acheive similar performance. those features were introduced to http in http/2 rfc7540. earlier versions of http are capable of conveying the semantic requirements of doh but would result in very poor performance for many uses cases." 45 {121, 59, 61, 47} {47: 0, 59: 1, 121: 2, 61: 2} {45: 0, 34: 1, 165: 2, 1: 2} "The messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, parallelism, priority, and header compression to acheive similar performance. Those features were introduced to HTTP in HTTP/2 RFC7540. Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but may result in very poor performance for many uses cases." "This text is rather subjective without data . Changing it to "" may "" still brings the point without needing a reference to a study proving the sentence true . ( whoops ; just realized I 've been blowing my pull requests by chaining them off a single branch ... my bad , and this will cause you some pain )" 90ffdb90666c2f5fefb94f3e2bad9303a98173a8 90ffdb90666c2f5fefb94f3e2bad9303a98173a8 draft-ietf-doh-dns-over-https-latest 70f4ea02178a1414d64a892cfeb64e15a2cfe2f82f412e43f153c229db2d2767 ['70f4ea02178a1414d64a892cfeb64e15a2cfe2f82f412e43f153c229db2d2767', '9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439'] B "the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, parallelism, priority, and header compression to acheive similar performance. those features were introduced to http in http/2 rfc7540. earlier versions of http are capable of conveying the semantic requirements of doh but may result in very poor performance for many uses cases." 44 {37, 79} {79: 0, 37: 1} {44: 0, 210: 1} "This is an example response for a query for the IN A records for ""www.example.com"" with recursion turned on. The response bears one record with an address of 93.184.216.34 and a TTL of 128 seconds." "Unfortunately , we should n't use the existing real - world address and should be using example ip space instead . unfortunately you included the PR changes in this PR too .. let 's see if the merge process can sort that out :) ok ! thanks ." 0e67da3d50441b9e5621d50d49a737cd27c765e6 0e67da3d50441b9e5621d50d49a737cd27c765e6 draft-ietf-doh-dns-over-https-latest 00caec7a8c5292c076576ae26e5a66aa63c70d03f024ed7580a9713f561f5f95 ['aab29aa5a84ed9e4b43de18963d935d378768f40bd644a678edfd8d929a7b8c3', '00caec7a8c5292c076576ae26e5a66aa63c70d03f024ed7580a9713f561f5f95'] A "this is an example response for a query for the in a records for ""www.example.com"" with recursion turned on. the response bears one record with an address of 93.184.216.34 and a ttl of 128 seconds." 44 {37, 79} {79: 0, 37: 1} {44: 0, 210: 1} "This is an example response for a query for the IN A records for ""www.example.com"" with recursion turned on. The response bears one record with an address of 192.0.2.1 and a TTL of 128 seconds." "Unfortunately , we should n't use the existing real - world address and should be using example ip space instead . unfortunately you included the PR changes in this PR too .. let 's see if the merge process can sort that out :) ok ! thanks ." 0e67da3d50441b9e5621d50d49a737cd27c765e6 0e67da3d50441b9e5621d50d49a737cd27c765e6 draft-ietf-doh-dns-over-https-latest aab29aa5a84ed9e4b43de18963d935d378768f40bd644a678edfd8d929a7b8c3 ['aab29aa5a84ed9e4b43de18963d935d378768f40bd644a678edfd8d929a7b8c3', '00caec7a8c5292c076576ae26e5a66aa63c70d03f024ed7580a9713f561f5f95'] B "this is an example response for a query for the in a records for ""www.example.com"" with recursion turned on. the response bears one record with an address of 192.0.2.1 and a ttl of 128 seconds." 43 {56, 83, 39} {56: 0, 39: 1, 83: 2} {43: 0, 76: 1, 41: 2} "This is technically already possible (since the server controls both the HTTP resources it exposes and the use of browser APIs by its content), but standardisation might make this easier to accomplish." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 7af67284870f2b35e06033e5a48a571312da0398d295bea82bf80b3ef1319248 ['7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', '7af67284870f2b35e06033e5a48a571312da0398d295bea82bf80b3ef1319248'] A "this is technically already possible (since the server controls both the http resources it exposes and the use of browser apis by its content), but standardisation might make this easier to accomplish." 43 {56, 83, 39} {56: 0, 39: 1, 83: 2} {43: 0, 76: 1, 41: 2} "This is technically already possible (since the server controls both the HTTP resources it exposes and the use of browser APIs by its content), but standardization might make this easier to accomplish." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206 ['7d60d3eb43a739eead60c4a0ff764eb7b5d0ff56b8b06368357d7dd35092b206', '7af67284870f2b35e06033e5a48a571312da0398d295bea82bf80b3ef1319248'] B "this is technically already possible (since the server controls both the http resources it exposes and the use of browser apis by its content), but standardization might make this easier to accomplish." 43 {92, 93} {92: 0, 93: 1} {43: 0, 1: 1} "The protocol must use a secure transport that meets the requirements for modern HTTPS." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest c7c2b1c39835e8b1c8b90efe7f1d83b71b87e390ab1a607b555ef8d1484bff2e ['c7c2b1c39835e8b1c8b90efe7f1d83b71b87e390ab1a607b555ef8d1484bff2e', '7fa7dd76c2673c7d35c5e9c11ea68b386960c5ad735b9254ab494e33a1458709'] A "the protocol must use a secure transport that meets the requirements for modern https." 43 {92, 93} {92: 0, 93: 1} {43: 0, 1: 1} "The protocol must use a secure transport that meets the requirements for HTTPS." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 7fa7dd76c2673c7d35c5e9c11ea68b386960c5ad735b9254ab494e33a1458709 ['c7c2b1c39835e8b1c8b90efe7f1d83b71b87e390ab1a607b555ef8d1484bff2e', '7fa7dd76c2673c7d35c5e9c11ea68b386960c5ad735b9254ab494e33a1458709'] B "the protocol must use a secure transport that meets the requirements for https." 43 {96, 22, 23, 58, 91} {23: 0, 96: 1, 58: 1, 22: 1, 91: 2} {43: 0, 64: 1, 19: 1, 60: 1, 80: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ct and a value indicating the media-format used for the body parameter. The value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. An empty value indicates the default application/dns-udpwireformat type (e.g. ct&body=...)." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] A "when using the get method the uri path must contain a query parameter with the name of ct and a value indicating the media-format used for the body parameter. the value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. an empty value indicates the default application/dns-udpwireformat type (e.g. ct&body=...)." 43 {96, 22, 23, 58, 91} {23: 0, 96: 1, 58: 1, 22: 1, 91: 2} {43: 0, 64: 1, 19: 1, 60: 1, 80: 2} "When using the GET method the URI path MUST contain a query parameter with the name of body. The value of the parameter is the content of the request encoded with base64url RFC4648. Using the GET method is friendlier to many HTTP cache implementations." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934 ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] A "when using the get method the uri path must contain a query parameter with the name of body. the value of the parameter is the content of the request encoded with base64url rfc4648. using the get method is friendlier to many http cache implementations." 43 {96, 22, 23, 58, 91} {23: 0, 96: 1, 58: 1, 22: 1, 91: 2} {43: 0, 64: 1, 19: 1, 60: 1, 80: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the body parameter. The value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. An empty value indicates the default application/dns- udpwireformat type (e.g. ct&body=...)." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] B "when using the get method the uri path must contain a query parameter with the name of ""ct"" and a value indicating the media-format used for the body parameter. the value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. an empty value indicates the default application/dns- udpwireformat type (e.g. ct&body=...)." 43 {96, 22, 23, 58, 91} {23: 0, 96: 1, 58: 1, 22: 1, 91: 2} {43: 0, 64: 1, 19: 1, 60: 1, 80: 2} "When using the GET method the URI path MUST contain a query parameter with the name of ""body"". The value of the parameter is the content of the request encoded with base64url RFC4648. Using the GET method is friendlier to many HTTP cache implementations." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4 ['339fc5d0b41f3dda74badce12874200865e232496ee2a625e67cbbfefd31c3d4', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'cdb3fd5be969465d2b93564327a183d48543e3b1429a20b91c1a3f68d6ea970e', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] B "when using the get method the uri path must contain a query parameter with the name of ""body"". the value of the parameter is the content of the request encoded with base64url rfc4648. using the get method is friendlier to many http cache implementations." 43 {38, 70, 19, 20, 118, 94} {94: 0, 19: 1, 20: 1, 118: 2, 70: 3, 38: 3} {43: 0, 78: 1, 1: 1, 142: 2, 144: 3, 122: 3} "At the time this is published, the response types are works in progress. The only known response type is ""application/dns- udpwireformat"", but it is possible that at least one JSON-based response format will be defined in the future." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest cac4379edb3ea1f1be5066e668a429264e4c6e1bb43cc1af2bfbcb4615557c94 ['cac4379edb3ea1f1be5066e668a429264e4c6e1bb43cc1af2bfbcb4615557c94', '1eb67f6f517e4fc148cf143742bc0bb28d91ed4cf0dc83dc4d13ee67a16d98e1'] A "at the time this is published, the response types are works in progress. the only known response type is ""application/dns- udpwireformat"", but it is possible that at least one json-based response format will be defined in the future." 43 {38, 70, 19, 20, 118, 94} {94: 0, 19: 1, 20: 1, 118: 2, 70: 3, 38: 3} {43: 0, 78: 1, 1: 1, 142: 2, 144: 3, 122: 3} "At the time this is published, the response types are works in progress. The only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that at least one JSON-based response format will be defined in the future." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 1eb67f6f517e4fc148cf143742bc0bb28d91ed4cf0dc83dc4d13ee67a16d98e1 ['cac4379edb3ea1f1be5066e668a429264e4c6e1bb43cc1af2bfbcb4615557c94', '1eb67f6f517e4fc148cf143742bc0bb28d91ed4cf0dc83dc4d13ee67a16d98e1'] B "at the time this is published, the response types are works in progress. the only response type defined in this document is ""application/dns-udpwireformat"", but it is possible that at least one json-based response format will be defined in the future." 43 {} {} {} "A DNS API Server SHOULD respond with HTTP status code 415 upon receiving a media type it is unable to process." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest f54ebf2494ca271baa96b414bd03cf9e83946d00c719668e4903a034157eead9 ['538d1aa865325044c56ff47318d83510fe19d311b265222f7ac5ad2a5ba686f2', 'f54ebf2494ca271baa96b414bd03cf9e83946d00c719668e4903a034157eead9'] A "a dns api server should respond with http status code 415 upon receiving a media type it is unable to process." 43 {} {} {} "A DNS API Server SHOULD respond with HTTP status code 415 (Unsupported Media Type) upon receiving a media type it is unable to process." "minor editoral changes thanks for these . The only one that 's a little iffy is the inclusion of the reason phrase for 415 ... but I think here its giving more explanatory value than an impression that particular phrase needs to be used ( which would n't jibe with http details ) ." e67528236b6fb9884f9ff99906db8aa4920ffcba e67528236b6fb9884f9ff99906db8aa4920ffcba draft-ietf-doh-dns-over-https-latest 538d1aa865325044c56ff47318d83510fe19d311b265222f7ac5ad2a5ba686f2 ['538d1aa865325044c56ff47318d83510fe19d311b265222f7ac5ad2a5ba686f2', 'f54ebf2494ca271baa96b414bd03cf9e83946d00c719668e4903a034157eead9'] B "a dns api server should respond with http status code 415 (unsupported media type) upon receiving a media type it is unable to process." 42 {33, 34, 18, 120, 26, 30} {120: 0, 34: 1, 33: 2, 26: 3, 18: 4, 30: 5} {42: 0, 21: 1, 33: 2, 39: 3, 56: 4, 101: 5} "Because DNS provides only caching but not revalidation semantics, DNS over HTTP responses should not carry revalidation response headers (such as Last-Modified: or Etag:) or return 304 responses." "The draft explicitly prohibits ETag , which suggests that we do n't care for revalidation requests at the HTTP layer ( and reliance on the resolver cache ) . I do n't think that 's necessary . Revalidation requests are a feature HTTP provides . It might be that the cost of serving revalidation is comparable to the cost of making a genuine request . DNS responses are generally modest in size . But using that as the basis of a decision here might foreclose on some use cases , for which it might be annoying to have to revise the protocol to support . Right now , I do n't see a concrete reason to prohibit revalidation , ETag , Last - Modified , and the works . If we consider the DNS API server as authoritative for its view of the state of particular resource records , then all these things can be used . For instance , Last - Modified can refer to the time that the DNS API server last received a response ( or even when it last received a response that changed its cached value , which is a different view on this ) . That said , the rough model I described in seems to suggest that DNS TTL limits what HTTP can do . If that model is correct , and both the DNS stack and API server cache right up to the TTL , then HTTP caching ca n't add any value . It does n't prohibit etag ( well , at least the author was n't trying to have it do that ) .. as you say , revalidation is a feature HTTP provides and its not cool for a foo over http draft to mess with that . Similar thing with 304 - 304 is n't real useful here and the draft is trying to steer you away from it in the name of quality of implementation .. but 304 is going to mean what HTTP says 304 means , doh does n't have the hubris to change that . there is a triggering non - normative ' should ' in the mix . it does try and explain that they are n't that helpful here - but I guess that 's really based on the pov that the DoH Server is a gateway to traditional DNS . In that scenario , which the draft talks about , a stale resource will probably revalidate because the traditional DNS side ca n't revalidate . I guess you could do some kind of byte for byte comparison and use 304 as a way to extend the ttl . and of course , there is no reason the DoH server ca n't just plain be the terminal authoritative server and thus easily do revalidation based on something other than the DNS protocol . based on those thoughts I 'll try and find a PR that downgrades this to "" FYI - writing revalidation code might not be that impactful "" feedback from singapore is to change to text so that it does n't discourage revalidation but instead explores the space a bit" f33323cbbb9c97b850d2a69433717b493e1da08c f33323cbbb9c97b850d2a69433717b493e1da08c draft-ietf-doh-dns-over-https-latest fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9 ['fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9'] A "because dns provides only caching but not revalidation semantics, dns over http responses should not carry revalidation response headers (such as last-modified: or etag:) or return 304 responses." 42 {123} {123: 0} {56: 0} "HTTP revalidation of cached DNS information may be of limited value as revalidation provides only a bandwidth benefit and DNS transactions are normally latency bound instead. Furthermore, the HTTP response headers that enable revalidation (such as ""Last- Modified"" and ""Etag"") are often fairly large when compared to the overall DNS response size, and have a variable nature that creates constant pressure on the HTTP/2 compression dictionary RFC7541. Other types of DNS data, such as zone transfers, may be larger and benefit more from revalidation. DNS API servers may wish to consider whether providing these optional response headers is worthwhile." "The draft explicitly prohibits ETag , which suggests that we do n't care for revalidation requests at the HTTP layer ( and reliance on the resolver cache ) . I do n't think that 's necessary . Revalidation requests are a feature HTTP provides . It might be that the cost of serving revalidation is comparable to the cost of making a genuine request . DNS responses are generally modest in size . But using that as the basis of a decision here might foreclose on some use cases , for which it might be annoying to have to revise the protocol to support . Right now , I do n't see a concrete reason to prohibit revalidation , ETag , Last - Modified , and the works . If we consider the DNS API server as authoritative for its view of the state of particular resource records , then all these things can be used . For instance , Last - Modified can refer to the time that the DNS API server last received a response ( or even when it last received a response that changed its cached value , which is a different view on this ) . That said , the rough model I described in seems to suggest that DNS TTL limits what HTTP can do . If that model is correct , and both the DNS stack and API server cache right up to the TTL , then HTTP caching ca n't add any value . It does n't prohibit etag ( well , at least the author was n't trying to have it do that ) .. as you say , revalidation is a feature HTTP provides and its not cool for a foo over http draft to mess with that . Similar thing with 304 - 304 is n't real useful here and the draft is trying to steer you away from it in the name of quality of implementation .. but 304 is going to mean what HTTP says 304 means , doh does n't have the hubris to change that . there is a triggering non - normative ' should ' in the mix . it does try and explain that they are n't that helpful here - but I guess that 's really based on the pov that the DoH Server is a gateway to traditional DNS . In that scenario , which the draft talks about , a stale resource will probably revalidate because the traditional DNS side ca n't revalidate . I guess you could do some kind of byte for byte comparison and use 304 as a way to extend the ttl . and of course , there is no reason the DoH server ca n't just plain be the terminal authoritative server and thus easily do revalidation based on something other than the DNS protocol . based on those thoughts I 'll try and find a PR that downgrades this to "" FYI - writing revalidation code might not be that impactful "" feedback from singapore is to change to text so that it does n't discourage revalidation but instead explores the space a bit" f33323cbbb9c97b850d2a69433717b493e1da08c f33323cbbb9c97b850d2a69433717b493e1da08c draft-ietf-doh-dns-over-https-latest fe8aedc5f6a1e18856a051a1333f36f564ad3111c214ffd53c62b62beb2af04f ['fe8aedc5f6a1e18856a051a1333f36f564ad3111c214ffd53c62b62beb2af04f'] B "http revalidation of cached dns information may be of limited value as revalidation provides only a bandwidth benefit and dns transactions are normally latency bound instead. furthermore, the http response headers that enable revalidation (such as ""last- modified"" and ""etag"") are often fairly large when compared to the overall dns response size, and have a variable nature that creates constant pressure on the http/2 compression dictionary rfc7541. other types of dns data, such as zone transfers, may be larger and benefit more from revalidation. dns api servers may wish to consider whether providing these optional response headers is worthwhile." 41 {56, 83, 39} {83: 0, 39: 1, 56: 2} {41: 0, 76: 1, 43: 2} "A secondary use case is allowing web applications to access DNS information, by using existing APIs in browsers to access it over HTTP." "As I understand it , the referenced CORS standard ( ) is obsolete , and has been superseded by the WHATWG fetch spec ( ) . Whether this is the official W3C position , I 'm not clear , but the WHATWG folks certainly seem to be running with the ball . any idea on what the right thing to do here is ? or ? The fetch spec is right , and referencing is the appropriate thing to do . This will make the IESG unhappy because fetch does n't meet their standards for stability of reference , but they will have to learn to deal with that ( there 's an open discussion about this point ) . The other problem is that this still does n't help provide the necessary context . Fetch is one of the most impenetrable and obtuse specifications in existence . Its description of CORS is really quite poor as an informational resource . It 's only really useful to someone who already knows what it means . I 'd recommend not worrying about that though . There are plenty of resources on the web that do a better job of explaining CORS . I would rather put a link to a readable reference than an unreadable one . Thus , I propose ( in all seriousness ) If the IESG does n't want that , they can propose something better . For an informative reference , that 's probably totally appropriate . I hope that we are n't going to start specifying the details of cross - origin requests . That would n't be appropriate here . I agree that the CORS portion of the fetch spec clear . Even the examples in the spec ( most of which I wrote ) do n't really help much , since the entire spec is really ' aimed ' at browser manufacturers rather than end - users . However , it has basically superseded the W3C spec originally referenced , and I 'm fed up with seeing references to the W3C spec ... It 's really no big deal , but I 'd prefer the link to refer to W3C / IETF hat on Early signs I 'm getting from W3C is that is the best thing to reference , although they 're still doing some process things on making that formal . My .02 - I think asking W3C / WebAppSec formally is going to be more productive than guessing or going on aesthetics . Setting a precedent of referencing based on our preferences rather than the responsible body 's is probably not something we want to do . I 'm happy to facilitate a discussion with them if the need arises . when I spoke to he said that getting W3C to update their CORS spec to note that it had been superseded by fetch was ... unlikely . Of course , this was a while back . I 'm more optimistic now . If you think this could be improved it would help if you filed actionable feedback as issues . ( FWIW , I 've had some success pointing Stack Overflow questions to the official standard . ) FWIW , I do n't think there was any intent by anyone to disparage the work you 've done on fetch . It 's just that it 's a standard that was documented from the POV of browser developers rather than end - users . At least , that 's my take . The "" CORS protocol "" section is not really meant for "" browser developers "" other than the "" HTTP new - header syntax "" section , which is normative for the header parsing requirements . Nothing else is normative there , all normative requirements are in the "" Fetching "" section . ( It 's also not for end users , but it 's certainly for web developers . )" 640e981f0a1c459316d19a317f29b3cc5a3d4efd 640e981f0a1c459316d19a317f29b3cc5a3d4efd draft-ietf-doh-dns-over-https-latest ccd737dac72d41d05f0bb0d2c912c50884dd203af0cf160b259c2d996ee9e617 ['bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', 'ccd737dac72d41d05f0bb0d2c912c50884dd203af0cf160b259c2d996ee9e617'] "https://www.w3.org/TR/cors/) https://fetch.spec.whatwg.org/). https://fetch.spec.whatwg.org/#http-cors-protocol https://en.wikipedia.org/wiki/Cross-originresourcesharing. https://fetch.spec.whatwg.org/#http-cors-protocol." A "a secondary use case is allowing web applications to access dns information, by using existing apis in browsers to access it over http." 41 {56, 83, 39} {83: 0, 39: 1, 56: 2} {41: 0, 76: 1, 43: 2} "A secondary use case is allowing web applications to access DNS information, by using existing APIs in browsers to access it over HTTP in a safe way consistent with Cross Origin Resource Sharing (CORS) CORS." "As I understand it , the referenced CORS standard ( ) is obsolete , and has been superseded by the WHATWG fetch spec ( ) . Whether this is the official W3C position , I 'm not clear , but the WHATWG folks certainly seem to be running with the ball . any idea on what the right thing to do here is ? or ? The fetch spec is right , and referencing is the appropriate thing to do . This will make the IESG unhappy because fetch does n't meet their standards for stability of reference , but they will have to learn to deal with that ( there 's an open discussion about this point ) . The other problem is that this still does n't help provide the necessary context . Fetch is one of the most impenetrable and obtuse specifications in existence . Its description of CORS is really quite poor as an informational resource . It 's only really useful to someone who already knows what it means . I 'd recommend not worrying about that though . There are plenty of resources on the web that do a better job of explaining CORS . I would rather put a link to a readable reference than an unreadable one . Thus , I propose ( in all seriousness ) If the IESG does n't want that , they can propose something better . For an informative reference , that 's probably totally appropriate . I hope that we are n't going to start specifying the details of cross - origin requests . That would n't be appropriate here . I agree that the CORS portion of the fetch spec clear . Even the examples in the spec ( most of which I wrote ) do n't really help much , since the entire spec is really ' aimed ' at browser manufacturers rather than end - users . However , it has basically superseded the W3C spec originally referenced , and I 'm fed up with seeing references to the W3C spec ... It 's really no big deal , but I 'd prefer the link to refer to W3C / IETF hat on Early signs I 'm getting from W3C is that is the best thing to reference , although they 're still doing some process things on making that formal . My .02 - I think asking W3C / WebAppSec formally is going to be more productive than guessing or going on aesthetics . Setting a precedent of referencing based on our preferences rather than the responsible body 's is probably not something we want to do . I 'm happy to facilitate a discussion with them if the need arises . when I spoke to he said that getting W3C to update their CORS spec to note that it had been superseded by fetch was ... unlikely . Of course , this was a while back . I 'm more optimistic now . If you think this could be improved it would help if you filed actionable feedback as issues . ( FWIW , I 've had some success pointing Stack Overflow questions to the official standard . ) FWIW , I do n't think there was any intent by anyone to disparage the work you 've done on fetch . It 's just that it 's a standard that was documented from the POV of browser developers rather than end - users . At least , that 's my take . The "" CORS protocol "" section is not really meant for "" browser developers "" other than the "" HTTP new - header syntax "" section , which is normative for the header parsing requirements . Nothing else is normative there , all normative requirements are in the "" Fetching "" section . ( It 's also not for end users , but it 's certainly for web developers . )" 640e981f0a1c459316d19a317f29b3cc5a3d4efd 640e981f0a1c459316d19a317f29b3cc5a3d4efd draft-ietf-doh-dns-over-https-latest bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8 ['bd14c4be18c0fef5e205b4e420722da26b8656a881c30c55ba7acda8f20b51b8', 'ccd737dac72d41d05f0bb0d2c912c50884dd203af0cf160b259c2d996ee9e617'] "https://www.w3.org/TR/cors/) https://fetch.spec.whatwg.org/). https://fetch.spec.whatwg.org/#http-cors-protocol https://en.wikipedia.org/wiki/Cross-originresourcesharing. https://fetch.spec.whatwg.org/#http-cors-protocol." B "a secondary use case is allowing web applications to access dns information, by using existing apis in browsers to access it over http in a safe way consistent with cross origin resource sharing (cors) cors." 40 {17, 2, 3} {17: 0, 3: 1, 2: 2} {40: 0, 119: 1, 129: 2} "Many HTTPS implementations perform real time third party checks of the revocation status of the certificates being used by TLS. If this check is done as part of the DNS API server connection procedure and the check itself requires DNS resolution to connect to the third party a deadlock can occur. The use of an OCSP RFC6960 server is one example of how this can happen. DNS API servers SHOULD utilize OCSP Stapling RFC6961 to provide the client with certificate revocation information that does not require contacting a third party." "In order to prevent a dependency on traditional DNS or a DNS query for CRL / OCSP certificate validation check , OCSP stapling should be used for the HTTPS web server . we 're probably better off describing the potential deadlock in revocation and mentioning stapling as a best practice .. but in general I think we want to stay away from specifying a whole lot of how - to - profile - https details as these things do differ between clients and we tend to accept all of the variants as "" "" .. for e.g." b0251ff2b1c1746d42d728ee8d5511926f304615 b0251ff2b1c1746d42d728ee8d5511926f304615 draft-ietf-doh-dns-over-https-latest 1d9ef1733bb8185e7f68e954808847031442cc43eb03659ac1967d02f4b498e4 ['1d9ef1733bb8185e7f68e954808847031442cc43eb03659ac1967d02f4b498e4'] B "many https implementations perform real time third party checks of the revocation status of the certificates being used by tls. if this check is done as part of the dns api server connection procedure and the check itself requires dns resolution to connect to the third party a deadlock can occur. the use of an ocsp rfc6960 server is one example of how this can happen. dns api servers should utilize ocsp stapling rfc6961 to provide the client with certificate revocation information that does not require contacting a third party." 39 {33, 34, 18, 120, 26, 30} {26: 0, 33: 1, 18: 1, 34: 2, 30: 2, 120: 3} {39: 0, 33: 1, 56: 1, 21: 2, 101: 2, 42: 3} "The Answer section of a DNS response contains one or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset is supposed to have the Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the Resource Record bearing the smallest TTL in the Answer section of the response. The HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "… during the singapore meeting tale noted that the http caching model , where 1 lifetime is assigned to the rrset is not necessarily the way a dns cache would handle things .. each rr can have its own ttl ( though one dns implementation does it this way ) . perhaps worth some explanatory text when talking about rrset granularity and http caches ." 686b0eab6f49ce4c687f92ac54e4fa18abc7aa5e 686b0eab6f49ce4c687f92ac54e4fa18abc7aa5e draft-ietf-doh-dns-over-https-latest 3b485b8d45b96aadf7710f1924f725e4bddad36666e9eface02931b25739c7b2 ['3b485b8d45b96aadf7710f1924f725e4bddad36666e9eface02931b25739c7b2', '25bbe1793ffe189c147ca76054675bf702f8cb9e8889d95ebc5b47e5e14d1fdd'] A "the answer section of a dns response contains one or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset is supposed to have the time to live (ttl) freshness information. different rrsets in the answer section can have different ttls. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the resource record bearing the smallest ttl in the answer section of the response. the http freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." 39 {33, 34, 18, 120, 26, 30} {26: 0, 33: 1, 18: 1, 34: 2, 30: 2, 120: 3} {39: 0, 33: 1, 56: 1, 21: 2, 101: 2, 42: 3} "The Answer section of a DNS response contains one or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset is supposed to have the Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs, though it is only possible for the HTTP response to have a single freshness lifetime. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the Resource Record bearing the smallest TTL in the Answer section of the response. The HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "… during the singapore meeting tale noted that the http caching model , where 1 lifetime is assigned to the rrset is not necessarily the way a dns cache would handle things .. each rr can have its own ttl ( though one dns implementation does it this way ) . perhaps worth some explanatory text when talking about rrset granularity and http caches ." 686b0eab6f49ce4c687f92ac54e4fa18abc7aa5e 686b0eab6f49ce4c687f92ac54e4fa18abc7aa5e draft-ietf-doh-dns-over-https-latest 25bbe1793ffe189c147ca76054675bf702f8cb9e8889d95ebc5b47e5e14d1fdd ['3b485b8d45b96aadf7710f1924f725e4bddad36666e9eface02931b25739c7b2', '25bbe1793ffe189c147ca76054675bf702f8cb9e8889d95ebc5b47e5e14d1fdd'] B "the answer section of a dns response contains one or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset is supposed to have the time to live (ttl) freshness information. different rrsets in the answer section can have different ttls, though it is only possible for the http response to have a single freshness lifetime. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the resource record bearing the smallest ttl in the answer section of the response. the http freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." 38 {} {} {} "The HTTPS connection provides transport security for the interaction between the DNS API server and client, but does not inherently ensure the authenticity of DNS data. A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server or it may choose to trust answers from a particular DNS API server, much as a DNS client might choose to trust answers from its recurvise DNS resolver." pr did some of this but was backed out . see that diff and discussion 0a21365d45ae0572d2aa078ea4fb74f547c77648 0a21365d45ae0572d2aa078ea4fb74f547c77648 draft-ietf-doh-dns-over-https-latest 4a41e621c1f9a914d830ec815943007577d3589c89849da2146d5bb810b5df26 ['4a41e621c1f9a914d830ec815943007577d3589c89849da2146d5bb810b5df26'] B "the https connection provides transport security for the interaction between the dns api server and client, but does not inherently ensure the authenticity of dns data. a dns api client may also perform full dnssec validation of answers received from a dns api server or it may choose to trust answers from a particular dns api server, much as a dns client might choose to trust answers from its recurvise dns resolver." 36 {8, 68, 69, 7} {69: 0, 68: 1, 7: 1, 8: 2} {36: 0, 1: 1, 34: 1, 99: 2} "DNS over HTTPS relies on the underlying HTTPS connection to provide confidentiality. By requiring at least RFC7540 levels of support for TLS, this protocol expects to use current best practices for confidential transport." Reverts dohwg / draft - ietf - doh - dns - over - https f92a046a9ad8f725b4f4ef9cd24d5538a2bad644 f92a046a9ad8f725b4f4ef9cd24d5538a2bad644 draft-ietf-doh-dns-over-https-latest 5209f26637cee37b6d01910e8924bfaf5c04696aa52f566f61daec9b0460d48b ['8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2', '5209f26637cee37b6d01910e8924bfaf5c04696aa52f566f61daec9b0460d48b'] A "dns over https relies on the underlying https connection to provide confidentiality. by requiring at least rfc7540 levels of support for tls, this protocol expects to use current best practices for confidential transport." 36 {8, 68, 69, 7} {69: 0, 68: 1, 7: 1, 8: 2} {36: 0, 1: 1, 34: 1, 99: 2} "Running DNS over HTTPS relies on the security of the underlying HTTP connection. By requiring at least RFC7540 levels of support for TLS, this protocol expects to use current best practices for secure transport." Reverts dohwg / draft - ietf - doh - dns - over - https f92a046a9ad8f725b4f4ef9cd24d5538a2bad644 f92a046a9ad8f725b4f4ef9cd24d5538a2bad644 draft-ietf-doh-dns-over-https-latest 8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2 ['8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2', '5209f26637cee37b6d01910e8924bfaf5c04696aa52f566f61daec9b0460d48b'] B "running dns over https relies on the security of the underlying http connection. by requiring at least rfc7540 levels of support for tls, this protocol expects to use current best practices for secure transport." 36 {104} {104: 0} {36: 0} "The HTTPS connection also authenticates the DNS API server but does not inherently assure of authenticity of DNS data. A DNS API client may choose to trust answers from a particular DNS API server, much as a DNS client might choose to trust answers from its recurvise DNS resolver. A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server." Reverts dohwg / draft - ietf - doh - dns - over - https f92a046a9ad8f725b4f4ef9cd24d5538a2bad644 f92a046a9ad8f725b4f4ef9cd24d5538a2bad644 draft-ietf-doh-dns-over-https-latest d568c8a6830fac093b575f3afa36ba2bf8e15f8e50c5429118113df9e0e1afc4 ['d568c8a6830fac093b575f3afa36ba2bf8e15f8e50c5429118113df9e0e1afc4'] A "the https connection also authenticates the dns api server but does not inherently assure of authenticity of dns data. a dns api client may choose to trust answers from a particular dns api server, much as a dns client might choose to trust answers from its recurvise dns resolver. a dns api client may also perform full dnssec validation of answers received from a dns api server." 35 {} {} {} "Joe Hildebrand contributed lots of material for a different iteration of this document. Helpful early comments were given by Ben Schwartz and Mark Nottingham." "recently I 've left the ack section mostly untouched until the final ( or near final ) edit . Easier to deal with it once . However I think its good idea to have leave the PR open and amend it as useful OK , having a "" living "" PR seems OK ." 000c9c8e0657162029a39025087131dbf03fb69b 000c9c8e0657162029a39025087131dbf03fb69b draft-ietf-doh-dns-over-https-latest 1240ca2b1044c4fc1acc14e74b7be7212303e5e49e3168c70e977d86f0dd0f9e ['1240ca2b1044c4fc1acc14e74b7be7212303e5e49e3168c70e977d86f0dd0f9e', 'b1517affddb8dec02b23e70673c4fa3f2c0e02551a6a288e4a9555feca072036'] A "joe hildebrand contributed lots of material for a different iteration of this document. helpful early comments were given by ben schwartz and mark nottingham." 35 {} {} {} "This work required a high level of cooperation between experts in different technologies. Thank you Ray Bellis, Stephane Bortzmeyer, Manu Bretelle, Tony Finch, Daniel Kahn Gilmor, Olafur Guomundsson, Wes Hardaker, Rory Hewitt, Joe Hildebrand, David Lawrence, Eliot Lear, John Mattson, Alex Mayrhofer, Mark Nottingham, Jim Reid, Adam Roach, Ben Schwartz, Davey Song, Daniel Stenberg, Andrew Sullivan, Martin Thomson, and Sam Weiler." "recently I 've left the ack section mostly untouched until the final ( or near final ) edit . Easier to deal with it once . However I think its good idea to have leave the PR open and amend it as useful OK , having a "" living "" PR seems OK ." 000c9c8e0657162029a39025087131dbf03fb69b 000c9c8e0657162029a39025087131dbf03fb69b draft-ietf-doh-dns-over-https-latest b1517affddb8dec02b23e70673c4fa3f2c0e02551a6a288e4a9555feca072036 ['1240ca2b1044c4fc1acc14e74b7be7212303e5e49e3168c70e977d86f0dd0f9e', 'b1517affddb8dec02b23e70673c4fa3f2c0e02551a6a288e4a9555feca072036'] B "this work required a high level of cooperation between experts in different technologies. thank you ray bellis, stephane bortzmeyer, manu bretelle, tony finch, daniel kahn gilmor, olafur guomundsson, wes hardaker, rory hewitt, joe hildebrand, david lawrence, eliot lear, john mattson, alex mayrhofer, mark nottingham, jim reid, adam roach, ben schwartz, davey song, daniel stenberg, andrew sullivan, martin thomson, and sam weiler." 34 {35, 78} {78: 0, 35: 1} {34: 0, 33: 1} "This document defines a specific protocol for sending DNS RFC1035 queries and getting DNS responses over modern versions of HTTP RFC7540 using https:// (and therefore TLS RFC5246 security for integrity and confidentiality). Each DNS query-response pair is mapped into a HTTP request-response pair." "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest a82ba5bdcaa55d13bce5d0716c0239f1188e8e679c2287246ebc650d04af6520 ['a82ba5bdcaa55d13bce5d0716c0239f1188e8e679c2287246ebc650d04af6520', '6ccc3f56aeabbfc140bc546ce4486f4df4136f68996da4faed47bf4867a86ec4'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" A "this document defines a specific protocol for sending dns rfc1035 queries and getting dns responses over modern versions of http rfc7540 using https:// (and therefore tls rfc5246 security for integrity and confidentiality). each dns query-response pair is mapped into a http request-response pair." 34 {35, 78} {78: 0, 35: 1} {34: 0, 33: 1} "This document defines a specific protocol for sending DNS RFC1035 queries and getting DNS responses over HTTP RFC7540 using https:// (and therefore TLS RFC5246 security for integrity and confidentiality). Each DNS query-response pair is mapped into a HTTP request-response pair." "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest 6ccc3f56aeabbfc140bc546ce4486f4df4136f68996da4faed47bf4867a86ec4 ['a82ba5bdcaa55d13bce5d0716c0239f1188e8e679c2287246ebc650d04af6520', '6ccc3f56aeabbfc140bc546ce4486f4df4136f68996da4faed47bf4867a86ec4'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" B "this document defines a specific protocol for sending dns rfc1035 queries and getting dns responses over http rfc7540 using https:// (and therefore tls rfc5246 security for integrity and confidentiality). each dns query-response pair is mapped into a http request-response pair." 34 {121, 59, 61, 47} {59: 0, 121: 1, 61: 1, 47: 1} {34: 0, 165: 1, 1: 1, 45: 1} "This protocol MUST use HTTP/2 RFC7540 or its successors in order to satisfy the security requirements of DNS over HTTPS. Further, the messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, priority, parallelism, and header compression, all of which are supported by HTTP/2 RFC7540 or its successors." "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest 826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8 ['826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8', 'fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154', '9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" A "this protocol must use http/2 rfc7540 or its successors in order to satisfy the security requirements of dns over https. further, the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, priority, parallelism, and header compression, all of which are supported by http/2 rfc7540 or its successors." 34 {121, 59, 61, 47} {59: 0, 121: 1, 61: 1, 47: 1} {34: 0, 165: 1, 1: 1, 45: 1} The minimum version of HTTP used by DOH SHOULD be HTTP/2 RFC7540. "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154 ['826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8', 'fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154', '9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" B the minimum version of http used by doh should be http/2 rfc7540. 34 {121, 59, 61, 47} {59: 0, 121: 1, 61: 1, 47: 1} {34: 0, 165: 1, 1: 1, 45: 1} "The messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, parallelism, priority, and header compression to acheive similar performance. Those features were introduced to HTTP in HTTP/2 RFC7540. Earlier versions of HTTP are capable of conveying the semantic requirements of DOH but would result in very poor performance for many uses cases." "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest 9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439 ['826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8', 'fab930f08aab528868e671bf3212aabf824ac220b717118c9a84bf4b2deb9154', '9aaba95f331a041d334b0bb7d05816d87204e0faeaa1880fd031b3aa4d16a439'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" B "the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, parallelism, priority, and header compression to acheive similar performance. those features were introduced to http in http/2 rfc7540. earlier versions of http are capable of conveying the semantic requirements of doh but would result in very poor performance for many uses cases." 34 {8, 68, 69, 7} {7: 0, 8: 1, 68: 1, 69: 1} {34: 0, 99: 1, 1: 1, 36: 1} "Running DNS over HTTPS relies on the security of the underlying HTTP connection. By requiring at least RFC7540 levels of support for TLS, this protocol expects to use current best practices for secure transport." "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest 8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2 ['0d3de8ab674456fbd0fb436cdb04209912d5fb844d8ff5a7b130d20fd46e024a', '8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" A "running dns over https relies on the security of the underlying http connection. by requiring at least rfc7540 levels of support for tls, this protocol expects to use current best practices for secure transport." 34 {8, 68, 69, 7} {7: 0, 8: 1, 68: 1, 69: 1} {34: 0, 99: 1, 1: 1, 36: 1} "Running DNS over HTTPS relies on the security of the underlying HTTP transport. Implementations utilizing HTTP/2 benefit from the TLS profile defined in RFC7540 Section 9.2." "I 'm ok with you merging this and closing 11 if you 're ok with it . I think that it is fine - or more precisely , invaluable - to explain what the semantics of server push are . But you do n't actually require HTTP/2 for this protocol to work . The only thing that requiring HTTP/2 gets you is tighter constraints on TLS usage . You can get those same constraints with some deployment recommendations . I would support strong recommendations that said HTTP/2 is useful for its multiplexing as well as similarly strong advice about following the advice in RFC 7525 . Those are both good practice . fairly clear feedback from singapore to change this to an endorsement ( SHOULD ) with better explanatory text focusing on performance and scale . What are the drawbacks of HTTP/2 ? Less readily available libraries ? more complex protocol ? Potential difficulties in integrating with current fronting stacks that may not be HTTP/2 ready ? On one hand , I feel like it may be easier to just enforce 1 stack for now rather than having to support multiple flavor of client / server protocols . A suboptimal HTTP/2 implementation may just be similar to HTTP/1 but at least we can try to move forward toward multiplexing . Making HTTP/2 a should may just leave us in limbo in HTTP/1 land . the singapore meeting is available here .. goto the 12 minute mark to review the discussion ( wrt your question about drawbacks ) . Text is in Last PR accepted . I suspect there will be a bit more wordsmithing , but this should cover the main issue well . thanks , I was in the room , but my notes were sparse and some of the cons I remembered of were more inconvenience to me than blockers . What I am not getting from the current spec , and which I believe touched on in , is that it is unclear whether or not a minimum HTTP version must be supported . Based on , do we consider that by supporting HTTP/1.1 we get HTTP/1.0 for free and consider 0.9 not an option ? Would n't not mandating a required version prevent client / server from having a known version to use ? Can we make this clearer maybe ? The decision was to rely only on HTTP semantics and not versions . So any HTTP implementation that is acceptable to its peer can be used ." 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f 968ac9dda7a8c053f2eae7dd8c08f79b71740e1f draft-ietf-doh-dns-over-https-latest 0d3de8ab674456fbd0fb436cdb04209912d5fb844d8ff5a7b130d20fd46e024a ['0d3de8ab674456fbd0fb436cdb04209912d5fb844d8ff5a7b130d20fd46e024a', '8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2'] "https://www.youtube.com/watch?v=1s1NDD92I&feature=youtu.be https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34 https://github.com/dohwg/draft-ietf-doh-dns-over-https/pull/34#discussion_r153363292, https://tools.ietf.org/html/rfc7230#appendix-A" B "running dns over https relies on the security of the underlying http transport. implementations utilizing http/2 benefit from the tls profile defined in rfc7540 section 9.2." 33 {35, 78} {35: 0, 78: 1} {33: 0, 34: 1} "This document defines a specific protocol for sending DNS RFC1035 queries and getting DNS responses over modern versions of HTTP RFC7540 using https:// (and therefore TLS RFC5246 security for integrity and confidentiality)." "… issue It is not stated crystal clear in the -01 draft if there 's only one DNS packet per HTTP request or if it is allowed to send more than one . Some wordings ( like ) suggest a single packet . I think it would benefit the document to be more explicit on this detail . good point - it is meant to be one , otherwise blocking and reordering ( which is resolved at the intra - request http level ) would be a problem . we 'll add text ." cba9d9bc91e446b4f5e8327d2d9773504119645c cba9d9bc91e446b4f5e8327d2d9773504119645c draft-ietf-doh-dns-over-https-latest 522c33b739486cba9a8335f1a85019025be466a85010258523316fa8fc87a2e6 ['522c33b739486cba9a8335f1a85019025be466a85010258523316fa8fc87a2e6', 'a82ba5bdcaa55d13bce5d0716c0239f1188e8e679c2287246ebc650d04af6520'] A "this document defines a specific protocol for sending dns rfc1035 queries and getting dns responses over modern versions of http rfc7540 using https:// (and therefore tls rfc5246 security for integrity and confidentiality)." 33 {35, 78} {35: 0, 78: 1} {33: 0, 34: 1} "This document defines a specific protocol for sending DNS RFC1035 queries and getting DNS responses over modern versions of HTTP RFC7540 using https:// (and therefore TLS RFC5246 security for integrity and confidentiality). Each DNS query-response pair is mapped into a HTTP request-response pair." "… issue It is not stated crystal clear in the -01 draft if there 's only one DNS packet per HTTP request or if it is allowed to send more than one . Some wordings ( like ) suggest a single packet . I think it would benefit the document to be more explicit on this detail . good point - it is meant to be one , otherwise blocking and reordering ( which is resolved at the intra - request http level ) would be a problem . we 'll add text ." cba9d9bc91e446b4f5e8327d2d9773504119645c cba9d9bc91e446b4f5e8327d2d9773504119645c draft-ietf-doh-dns-over-https-latest a82ba5bdcaa55d13bce5d0716c0239f1188e8e679c2287246ebc650d04af6520 ['522c33b739486cba9a8335f1a85019025be466a85010258523316fa8fc87a2e6', 'a82ba5bdcaa55d13bce5d0716c0239f1188e8e679c2287246ebc650d04af6520'] B "this document defines a specific protocol for sending dns rfc1035 queries and getting dns responses over modern versions of http rfc7540 using https:// (and therefore tls rfc5246 security for integrity and confidentiality). each dns query-response pair is mapped into a http request-response pair." 33 {65, 67, 36, 73, 10, 75, 113} {75: 0, 65: 1, 73: 2, 10: 2, 67: 3, 36: 3, 113: 3} {33: 0, 61: 1, 131: 2, 6: 2, 133: 3, 128: 3, 199: 3} "A DNS API Client encodes the DNS query into the HTTP request using either the HTTP GET or POST methods." "… issue It is not stated crystal clear in the -01 draft if there 's only one DNS packet per HTTP request or if it is allowed to send more than one . Some wordings ( like ) suggest a single packet . I think it would benefit the document to be more explicit on this detail . good point - it is meant to be one , otherwise blocking and reordering ( which is resolved at the intra - request http level ) would be a problem . we 'll add text ." cba9d9bc91e446b4f5e8327d2d9773504119645c cba9d9bc91e446b4f5e8327d2d9773504119645c draft-ietf-doh-dns-over-https-latest a33177ce364b4705c20229eb2aa40ba1c357b80ba5294dd5d42c3127d01626ea ['a33177ce364b4705c20229eb2aa40ba1c357b80ba5294dd5d42c3127d01626ea', '8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e'] A "a dns api client encodes the dns query into the http request using either the http get or post methods." 33 {65, 67, 36, 73, 10, 75, 113} {75: 0, 65: 1, 73: 2, 10: 2, 67: 3, 36: 3, 113: 3} {33: 0, 61: 1, 131: 2, 6: 2, 133: 3, 128: 3, 199: 3} "A DNS API Client encodes a single DNS query into the HTTP request using either the HTTP GET or POST methods." "… issue It is not stated crystal clear in the -01 draft if there 's only one DNS packet per HTTP request or if it is allowed to send more than one . Some wordings ( like ) suggest a single packet . I think it would benefit the document to be more explicit on this detail . good point - it is meant to be one , otherwise blocking and reordering ( which is resolved at the intra - request http level ) would be a problem . we 'll add text ." cba9d9bc91e446b4f5e8327d2d9773504119645c cba9d9bc91e446b4f5e8327d2d9773504119645c draft-ietf-doh-dns-over-https-latest 8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e ['a33177ce364b4705c20229eb2aa40ba1c357b80ba5294dd5d42c3127d01626ea', '8a9eb7b502c9cfe4a51c66dca6003868d0e8882b1e67e0a468d0a32930b2517e'] B "a dns api client encodes a single dns query into the http request using either the http get or post methods." 33 {33, 34, 18, 120, 26, 30} {33: 0, 26: 1, 34: 1, 18: 2, 120: 2, 30: 3} {33: 0, 39: 1, 21: 1, 56: 2, 42: 2, 101: 3} "A DNS response may contain a set of DNS Resource Records each with its own Time To Live (TTL) freshness information. The freshness lifetime (RFC7234 Section 4.2) of the HTTP response should be coordinated with the Resource Record bearing the smallest TTL in the response. The freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "… issue It is not stated crystal clear in the -01 draft if there 's only one DNS packet per HTTP request or if it is allowed to send more than one . Some wordings ( like ) suggest a single packet . I think it would benefit the document to be more explicit on this detail . good point - it is meant to be one , otherwise blocking and reordering ( which is resolved at the intra - request http level ) would be a problem . we 'll add text ." cba9d9bc91e446b4f5e8327d2d9773504119645c cba9d9bc91e446b4f5e8327d2d9773504119645c draft-ietf-doh-dns-over-https-latest 4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713 ['4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713', '3b485b8d45b96aadf7710f1924f725e4bddad36666e9eface02931b25739c7b2'] A "a dns response may contain a set of dns resource records each with its own time to live (ttl) freshness information. the freshness lifetime (rfc7234 section 4.2) of the http response should be coordinated with the resource record bearing the smallest ttl in the response. the freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." 33 {33, 34, 18, 120, 26, 30} {33: 0, 26: 1, 34: 1, 18: 2, 120: 2, 30: 3} {33: 0, 39: 1, 21: 1, 56: 2, 42: 2, 101: 3} "The Answer section of a DNS response contains one or more RRsets. (RRsets are defined in RFC7719.) According to RFC2181, each resource record in an RRset is supposed to have the Time To Live (TTL) freshness information. Different RRsets in the Answer section can have different TTLs. The HTTP response freshness lifetime (RFC7234 Section 4.2) should be coordinated with the Resource Record bearing the smallest TTL in the Answer section of the response. The HTTP freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "… issue It is not stated crystal clear in the -01 draft if there 's only one DNS packet per HTTP request or if it is allowed to send more than one . Some wordings ( like ) suggest a single packet . I think it would benefit the document to be more explicit on this detail . good point - it is meant to be one , otherwise blocking and reordering ( which is resolved at the intra - request http level ) would be a problem . we 'll add text ." cba9d9bc91e446b4f5e8327d2d9773504119645c cba9d9bc91e446b4f5e8327d2d9773504119645c draft-ietf-doh-dns-over-https-latest 3b485b8d45b96aadf7710f1924f725e4bddad36666e9eface02931b25739c7b2 ['4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713', '3b485b8d45b96aadf7710f1924f725e4bddad36666e9eface02931b25739c7b2'] B "the answer section of a dns response contains one or more rrsets. (rrsets are defined in rfc7719.) according to rfc2181, each resource record in an rrset is supposed to have the time to live (ttl) freshness information. different rrsets in the answer section can have different ttls. the http response freshness lifetime (rfc7234 section 4.2) should be coordinated with the resource record bearing the smallest ttl in the answer section of the response. the http freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." 25 {8, 68, 69, 7} {69: 0, 68: 1, 7: 1, 8: 2} {36: 0, 1: 1, 34: 1, 99: 2} "Running DNS over HTTPS relies on the security of the underlying HTTP connection. By requiring at least RFC7540 levels of support for TLS, this protocol expects to use current best practices for secure transport." "Separate out confidentiality and authentication - -01 said "" Running DNS over HTTPS relies on the security of the underlying HTTP connection . "" This PR clarifies that the HTTPS connection can only provide confidentiality and authentication of the DNS API server . Clients wanting authentic DNS answers need to perform DNSSEC validation or explicitly trust the DNS API server . I reverted this because my comment above was not satisfied before merge . pr did some of this but was backed out . see that diff and discussion I agree with the notion of separating transport security from e2e validation .. but I think the changes overshoot a touch . I would leave the first paragraph alone - https provides more than confidentiality and I do n't think the text suggests anything other than doh relies on https . I like your second paragraph - though if we leave the first one in tact you probably want to remove the "" also authenticates "" part .. it can be recast as transport security between api server and client vs e2e validation" c1f4d2a95f56e99ef28ab4ff4483a9ece08b3f82 c1f4d2a95f56e99ef28ab4ff4483a9ece08b3f82 draft-ietf-doh-dns-over-https-latest 8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2 ['8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2', '5209f26637cee37b6d01910e8924bfaf5c04696aa52f566f61daec9b0460d48b'] A "running dns over https relies on the security of the underlying http connection. by requiring at least rfc7540 levels of support for tls, this protocol expects to use current best practices for secure transport." 25 {8, 68, 69, 7} {69: 0, 68: 1, 7: 1, 8: 2} {36: 0, 1: 1, 34: 1, 99: 2} "DNS over HTTPS relies on the underlying HTTPS connection to provide confidentiality. By requiring at least RFC7540 levels of support for TLS, this protocol expects to use current best practices for confidential transport." "Separate out confidentiality and authentication - -01 said "" Running DNS over HTTPS relies on the security of the underlying HTTP connection . "" This PR clarifies that the HTTPS connection can only provide confidentiality and authentication of the DNS API server . Clients wanting authentic DNS answers need to perform DNSSEC validation or explicitly trust the DNS API server . I reverted this because my comment above was not satisfied before merge . pr did some of this but was backed out . see that diff and discussion I agree with the notion of separating transport security from e2e validation .. but I think the changes overshoot a touch . I would leave the first paragraph alone - https provides more than confidentiality and I do n't think the text suggests anything other than doh relies on https . I like your second paragraph - though if we leave the first one in tact you probably want to remove the "" also authenticates "" part .. it can be recast as transport security between api server and client vs e2e validation" c1f4d2a95f56e99ef28ab4ff4483a9ece08b3f82 c1f4d2a95f56e99ef28ab4ff4483a9ece08b3f82 draft-ietf-doh-dns-over-https-latest 5209f26637cee37b6d01910e8924bfaf5c04696aa52f566f61daec9b0460d48b ['8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2', '5209f26637cee37b6d01910e8924bfaf5c04696aa52f566f61daec9b0460d48b'] B "dns over https relies on the underlying https connection to provide confidentiality. by requiring at least rfc7540 levels of support for tls, this protocol expects to use current best practices for confidential transport." 25 {104} {104: 0} {36: 0} "The HTTPS connection also authenticates the DNS API server but does not inherently assure of authenticity of DNS data. A DNS API client may choose to trust answers from a particular DNS API server, much as a DNS client might choose to trust answers from its recurvise DNS resolver. A DNS API client may also perform full DNSSEC validation of answers received from a DNS API server." "Separate out confidentiality and authentication - -01 said "" Running DNS over HTTPS relies on the security of the underlying HTTP connection . "" This PR clarifies that the HTTPS connection can only provide confidentiality and authentication of the DNS API server . Clients wanting authentic DNS answers need to perform DNSSEC validation or explicitly trust the DNS API server . I reverted this because my comment above was not satisfied before merge . pr did some of this but was backed out . see that diff and discussion I agree with the notion of separating transport security from e2e validation .. but I think the changes overshoot a touch . I would leave the first paragraph alone - https provides more than confidentiality and I do n't think the text suggests anything other than doh relies on https . I like your second paragraph - though if we leave the first one in tact you probably want to remove the "" also authenticates "" part .. it can be recast as transport security between api server and client vs e2e validation" c1f4d2a95f56e99ef28ab4ff4483a9ece08b3f82 c1f4d2a95f56e99ef28ab4ff4483a9ece08b3f82 draft-ietf-doh-dns-over-https-latest d568c8a6830fac093b575f3afa36ba2bf8e15f8e50c5429118113df9e0e1afc4 ['d568c8a6830fac093b575f3afa36ba2bf8e15f8e50c5429118113df9e0e1afc4'] B "the https connection also authenticates the dns api server but does not inherently assure of authenticity of dns data. a dns api client may choose to trust answers from a particular dns api server, much as a dns client might choose to trust answers from its recurvise dns resolver. a dns api client may also perform full dnssec validation of answers received from a dns api server." 21 {33, 34, 18, 120, 26, 30} {34: 0, 120: 1, 33: 1, 26: 2, 18: 3, 30: 4} {21: 0, 42: 1, 33: 1, 39: 2, 56: 3, 101: 4} "In the HTTP responses, the HTTP cache headers SHOULD be set to expire at the same time as the shortest DNS TTL in the response. Because DNS provides only caching but not revalidation semantics, DNS over HTTP responses should not carry revalidation response headers (such as Last-Modified: or Etag:) or return 304 responses." "closes issue thanks . The draft says "" the HTTP cache headers SHOULD be set to expire at the same time as the shortest DNS TTL in the response "" , which might not be strong enough . A SHOULD might be OK for shorter , but allowing the value to increase might be unwise , depending on the caching model . agree Seems like it should be MUST : some typos , also , I ca n't seem to parse 1 of the sentence ." e076318264c435ad15d319c39d433f3f5391e6a4 e076318264c435ad15d319c39d433f3f5391e6a4 draft-ietf-doh-dns-over-https-latest 3c200f9a267f00fd6844a910dac42b5fb5a3923d239d74f29d9fd7c5f1583a31 ['4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713', 'f4e0fa1c3d2560d97e1950816e839eccc78ba460503832f15ad3c1e52a38a25c', '3c200f9a267f00fd6844a910dac42b5fb5a3923d239d74f29d9fd7c5f1583a31', 'fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9'] A "in the http responses, the http cache headers should be set to expire at the same time as the shortest dns ttl in the response. because dns provides only caching but not revalidation semantics, dns over http responses should not carry revalidation response headers (such as last-modified: or etag:) or return 304 responses." 21 {33, 34, 18, 120, 26, 30} {34: 0, 120: 1, 33: 1, 26: 2, 18: 3, 30: 4} {21: 0, 42: 1, 33: 1, 39: 2, 56: 3, 101: 4} "A DNS response may contain a set of DNS Resource Records each with its own Time To Live (TTL) freshness information. The freshness lifetime (RFC7234 Section 4.2) of the HTTP response should be coordinated with the Resource Record bearing the smallest TTL in the response. The freshness lifetime SHOULD be set to expire at the same time any of the DNS Records reach a 0 TTL. The response freshness lifetime MUST NOT be greater than that indicated by the DNS Record with the smallest TTL in the response." "closes issue thanks . The draft says "" the HTTP cache headers SHOULD be set to expire at the same time as the shortest DNS TTL in the response "" , which might not be strong enough . A SHOULD might be OK for shorter , but allowing the value to increase might be unwise , depending on the caching model . agree Seems like it should be MUST : some typos , also , I ca n't seem to parse 1 of the sentence ." e076318264c435ad15d319c39d433f3f5391e6a4 e076318264c435ad15d319c39d433f3f5391e6a4 draft-ietf-doh-dns-over-https-latest 4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713 ['4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713', 'f4e0fa1c3d2560d97e1950816e839eccc78ba460503832f15ad3c1e52a38a25c', '3c200f9a267f00fd6844a910dac42b5fb5a3923d239d74f29d9fd7c5f1583a31', 'fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9'] B "a dns response may contain a set of dns resource records each with its own time to live (ttl) freshness information. the freshness lifetime (rfc7234 section 4.2) of the http response should be coordinated with the resource record bearing the smallest ttl in the response. the freshness lifetime should be set to expire at the same time any of the dns records reach a 0 ttl. the response freshness lifetime must not be greater than that indicated by the dns record with the smallest ttl in the response." 21 {33, 34, 18, 120, 26, 30} {34: 0, 120: 1, 33: 1, 26: 2, 18: 3, 30: 4} {21: 0, 42: 1, 33: 1, 39: 2, 56: 3, 101: 4} "A DNS API Client that receives a response without an explicit freshness lifetime MUST NOT assign that response a heuristic freshness (RFC7234 Section 4.2.2.) greater than that indicated by the DNS Record with the smallest TTL in the response." "closes issue thanks . The draft says "" the HTTP cache headers SHOULD be set to expire at the same time as the shortest DNS TTL in the response "" , which might not be strong enough . A SHOULD might be OK for shorter , but allowing the value to increase might be unwise , depending on the caching model . agree Seems like it should be MUST : some typos , also , I ca n't seem to parse 1 of the sentence ." e076318264c435ad15d319c39d433f3f5391e6a4 e076318264c435ad15d319c39d433f3f5391e6a4 draft-ietf-doh-dns-over-https-latest f4e0fa1c3d2560d97e1950816e839eccc78ba460503832f15ad3c1e52a38a25c ['4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713', 'f4e0fa1c3d2560d97e1950816e839eccc78ba460503832f15ad3c1e52a38a25c', '3c200f9a267f00fd6844a910dac42b5fb5a3923d239d74f29d9fd7c5f1583a31', 'fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9'] B "a dns api client that receives a response without an explicit freshness lifetime must not assign that response a heuristic freshness (rfc7234 section 4.2.2.) greater than that indicated by the dns record with the smallest ttl in the response." 21 {33, 34, 18, 120, 26, 30} {34: 0, 120: 1, 33: 1, 26: 2, 18: 3, 30: 4} {21: 0, 42: 1, 33: 1, 39: 2, 56: 3, 101: 4} "Because DNS provides only caching but not revalidation semantics, DNS over HTTP responses should not carry revalidation response headers (such as Last-Modified: or Etag:) or return 304 responses." "closes issue thanks . The draft says "" the HTTP cache headers SHOULD be set to expire at the same time as the shortest DNS TTL in the response "" , which might not be strong enough . A SHOULD might be OK for shorter , but allowing the value to increase might be unwise , depending on the caching model . agree Seems like it should be MUST : some typos , also , I ca n't seem to parse 1 of the sentence ." e076318264c435ad15d319c39d433f3f5391e6a4 e076318264c435ad15d319c39d433f3f5391e6a4 draft-ietf-doh-dns-over-https-latest fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9 ['4d6ad88ac5c7dcd944d8785761053cae40c797020afc80e92c96ec0904860713', 'f4e0fa1c3d2560d97e1950816e839eccc78ba460503832f15ad3c1e52a38a25c', '3c200f9a267f00fd6844a910dac42b5fb5a3923d239d74f29d9fd7c5f1583a31', 'fa8f1fe564eae8aae4db96a186289b16d972ead36fed2d177eed4714a6b4a4f9'] B "because dns provides only caching but not revalidation semantics, dns over http responses should not carry revalidation response headers (such as last-modified: or etag:) or return 304 responses." 20 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {12: 0, 64: 1, 5: 2, 4: 3, 54: 3, 98: 4, 53: 4, 99: 5, 107: 5, 106: 6} {20: 0, 61: 1, 65: 2, 78: 3, 131: 3, 142: 4, 199: 4, 144: 5} "For example, assume a DNS API server is following this specification on origin https://dnsserver.example.net/ and the well-known path. The DNS API client chooses to send its requests in application/dns- udpwirefomat but indicates it can parse replies in that format or as a JSON-based content type." "proposed to close issue thanks If it is n't then including it in an example might be unwise . Explicitly pointing out that it is currently fictitious might avoid confusion . some typos" 43cc6341c92d334051d42808d9e97a3abae49b4d 43cc6341c92d334051d42808d9e97a3abae49b4d draft-ietf-doh-dns-over-https-latest e06eba6e6dd68fe5c26e271b6b72d0d46ad7eae37daee8b07c594f6d014ba63b ['1727b3c9971f01f76457c93094c768678800756be9e1e3ce34d57c71f0c0045f', '4911e00eae5163a05b56250c5d0bcfc8a08e5d2875020b79aa437d9b4f124c89', '8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2', 'e06eba6e6dd68fe5c26e271b6b72d0d46ad7eae37daee8b07c594f6d014ba63b'] A "for example, assume a dns api server is following this specification on origin https://dnsserver.example.net/ and the well-known path. the dns api client chooses to send its requests in application/dns- udpwirefomat but indicates it can parse replies in that format or as a json-based content type." 20 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {12: 0, 64: 1, 5: 2, 4: 3, 54: 3, 98: 4, 53: 4, 99: 5, 107: 5, 106: 6} {20: 0, 61: 1, 65: 2, 78: 3, 131: 3, 142: 4, 199: 4, 144: 5} The examples uses HTTP/2 formatting from RFC7540. "proposed to close issue thanks If it is n't then including it in an example might be unwise . Explicitly pointing out that it is currently fictitious might avoid confusion . some typos" 43cc6341c92d334051d42808d9e97a3abae49b4d 43cc6341c92d334051d42808d9e97a3abae49b4d draft-ietf-doh-dns-over-https-latest 1727b3c9971f01f76457c93094c768678800756be9e1e3ce34d57c71f0c0045f ['1727b3c9971f01f76457c93094c768678800756be9e1e3ce34d57c71f0c0045f', '4911e00eae5163a05b56250c5d0bcfc8a08e5d2875020b79aa437d9b4f124c89', '8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2', 'e06eba6e6dd68fe5c26e271b6b72d0d46ad7eae37daee8b07c594f6d014ba63b'] A the examples uses http/2 formatting from rfc7540. 20 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {12: 0, 64: 1, 5: 2, 4: 3, 54: 3, 98: 4, 53: 4, 99: 5, 107: 5, 106: 6} {20: 0, 61: 1, 65: 2, 78: 3, 131: 3, 142: 4, 199: 4, 144: 5} These examples use HTTP/2 style formatting from RFC7540. "proposed to close issue thanks If it is n't then including it in an example might be unwise . Explicitly pointing out that it is currently fictitious might avoid confusion . some typos" 43cc6341c92d334051d42808d9e97a3abae49b4d 43cc6341c92d334051d42808d9e97a3abae49b4d draft-ietf-doh-dns-over-https-latest 4911e00eae5163a05b56250c5d0bcfc8a08e5d2875020b79aa437d9b4f124c89 ['1727b3c9971f01f76457c93094c768678800756be9e1e3ce34d57c71f0c0045f', '4911e00eae5163a05b56250c5d0bcfc8a08e5d2875020b79aa437d9b4f124c89', '8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2', 'e06eba6e6dd68fe5c26e271b6b72d0d46ad7eae37daee8b07c594f6d014ba63b'] B these examples use http/2 style formatting from rfc7540. 20 {64, 98, 99, 4, 5, 106, 107, 12, 53, 54} {12: 0, 64: 1, 5: 2, 4: 3, 54: 3, 98: 4, 53: 4, 99: 5, 107: 5, 106: 6} {20: 0, 61: 1, 65: 2, 78: 3, 131: 3, 142: 4, 199: 4, 144: 5} "For this example assume a DNS API server is following this specification on origin https://dnsserver.example.net/ and the well- known path. The DNS API client chooses to send its requests in application/dns-udpwirefomat but indicates it can parse replies in that format or as a hypothetical JSON-based content type. The application/simpledns+json type used by this example is currently fictitious." "proposed to close issue thanks If it is n't then including it in an example might be unwise . Explicitly pointing out that it is currently fictitious might avoid confusion . some typos" 43cc6341c92d334051d42808d9e97a3abae49b4d 43cc6341c92d334051d42808d9e97a3abae49b4d draft-ietf-doh-dns-over-https-latest 8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2 ['1727b3c9971f01f76457c93094c768678800756be9e1e3ce34d57c71f0c0045f', '4911e00eae5163a05b56250c5d0bcfc8a08e5d2875020b79aa437d9b4f124c89', '8a037fd00e5d18216c1fe9eba0c3375f91279c16422819de9aa0b1591bf280d2', 'e06eba6e6dd68fe5c26e271b6b72d0d46ad7eae37daee8b07c594f6d014ba63b'] B "for this example assume a dns api server is following this specification on origin https://dnsserver.example.net/ and the well- known path. the dns api client chooses to send its requests in application/dns-udpwirefomat but indicates it can parse replies in that format or as a hypothetical json-based content type. the application/simpledns+json type used by this example is currently fictitious." 19 {96, 22, 23, 58, 91} {58: 0, 23: 1, 96: 2, 22: 2, 91: 3} {19: 0, 43: 1, 64: 2, 60: 2, 80: 3} "When using the POST method, the DNS query is included as the message body of the HTTP request and the Content-Type request header indicates the media type of the message. POST-ed requests are smaller than their GET equivalents." "closes issue I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a ct name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ct ( what if you mis - spell it ? ) Actually what was the reasoning behind passing Content - Type as a query parameter instead of as the Content - Type header ? Presumably any DNS API server must be able to parse headers ... Content - Type as a header refers to the message payload which is n't what we 're trying to describe here .. ( its used on POST , where it matches .. ) referring back to pull , you wrote this : "" I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ( what if you mis - spell it ? ) "" I 'm fine with having now named , but I 'm concerned by the use of a the query parameter without a value at all ... Whilst this is probably too unwieldy : /dns - query?ct = application / dns - udpwireformat&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB the use of alone just /dns - query?ct&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB Yes , I know it 's valid according to RFC 3986 , but I 'm wondering whether , since it 's very unlikely that a JSON format would be passed using a GET , we could treat the absence of as implying ? Alternatively we could allow a couple of special values for the parameter : - implies that the format is - implies that the format is There 's a potential different approach here , use a specific parameter name for each type of query . That is , has a value that is json= ... application / dns - query+json ` and so on . That way you can have multiple formats on the request ( for backwards compatibility ) . The only downside is that the backwardly compatible request will cause a hit on cache coherency , so the server might want to redirect to a canonical URL if it is concerned that the client wo n't consistently include both formats . ( That redirect can be followed by a push if you are concerned about round trips . ) This feels cache - unfriendly to me . What is the problem with the default for ct ? It 's a tiny hit to cache unfriendliness ( you do n't get coherency between clients that inconsistently construct requests , or you have to redirect - and - push ) , agreed . But it gets you the agility you wanted . Oh , forgot to point out that this tends to be self - correcting . There are usually only a few different configurations of clients , so while those that support two request formats wo n't necessarily share cache hits with those that only support one , those two populations will probably happily share a cache . So to be clear , you 'd allow requests like any of the following : Personally I like this - clear and neat . Maybe I 'd change the parameter to ( so the two options are and , but that 's just me . It removes the necessity to specify the type separately and it 's ' neat ' . Plus extensible for future formats . Yeah . Pick whatever colour you like for the shed , I 'm not planning to live in it . describes a method that is more flexible again in . A better solution in my opinion ( if you set the second half of that issue aside for the moment ) . I think the color of the shed is fine for this document , and if discovery would like to do some fancy templating to enable rainbows that 's fine for that separate effort . could be much shorter . Bytes in requests matter and these are going to be used in critical conditions ( early in the connection where the number of requests you can make depend on how big they are ) . These wo n't compress well enough with HPACK to get any real benefit . A short name would be an improvement . A default value for would be even better . Could go the alt - svc approach ( ) and use two - letter combos , e.g. "" ct "" for , but really there out to be a standard for that approach ... I like the change to ' ct ' for Content - Type , but I think ' body ' is important enough where reducing it by only two bytes to ' rb ' is n't worth the change . Perhaps there could be an abbreviated ' ct ' value - e.g. ' ct = du ' ? Or , if ' ct ' is not specified , then ' application / dns - udpwireformat ' iMUST be assumed by the DNS server ?" fb31cda5950400c93017fc49096d1f7e4cf183e0 fb31cda5950400c93017fc49096d1f7e4cf183e0 draft-ietf-doh-dns-over-https-latest b8ec7d5063ce550170279530749a83e23908984958cd794f9589041bb8c7d285 ['804683cd1896a3f22e9aed540302b920f4e0acb5185ae135c58d1d1276579335', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'b8ec7d5063ce550170279530749a83e23908984958cd794f9589041bb8c7d285', 'd2d81b62aa35c9d05340f4e311651515e578d7855f07f7839e34923422e284dd', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] https://tools.ietf.org/html/draft-ietf-httpbis-alt-svc-14#section-3.1) A "when using the post method, the dns query is included as the message body of the http request and the content-type request header indicates the media type of the message. post-ed requests are smaller than their get equivalents." 19 {96, 22, 23, 58, 91} {58: 0, 23: 1, 96: 2, 22: 2, 91: 3} {19: 0, 43: 1, 64: 2, 60: 2, 80: 3} "When using the GET method, the URI path MUST contain a query parameter of the form content-type=TTT and another of the form body=BBBB, where ""TTT"" is the media type of the format used for the body parameter, and ""BBB"" is the content of the body encoded with base64url RFC4648. Using the GET method is friendlier to many HTTP cache implementations." "closes issue I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a ct name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ct ( what if you mis - spell it ? ) Actually what was the reasoning behind passing Content - Type as a query parameter instead of as the Content - Type header ? Presumably any DNS API server must be able to parse headers ... Content - Type as a header refers to the message payload which is n't what we 're trying to describe here .. ( its used on POST , where it matches .. ) referring back to pull , you wrote this : "" I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ( what if you mis - spell it ? ) "" I 'm fine with having now named , but I 'm concerned by the use of a the query parameter without a value at all ... Whilst this is probably too unwieldy : /dns - query?ct = application / dns - udpwireformat&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB the use of alone just /dns - query?ct&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB Yes , I know it 's valid according to RFC 3986 , but I 'm wondering whether , since it 's very unlikely that a JSON format would be passed using a GET , we could treat the absence of as implying ? Alternatively we could allow a couple of special values for the parameter : - implies that the format is - implies that the format is There 's a potential different approach here , use a specific parameter name for each type of query . That is , has a value that is json= ... application / dns - query+json ` and so on . That way you can have multiple formats on the request ( for backwards compatibility ) . The only downside is that the backwardly compatible request will cause a hit on cache coherency , so the server might want to redirect to a canonical URL if it is concerned that the client wo n't consistently include both formats . ( That redirect can be followed by a push if you are concerned about round trips . ) This feels cache - unfriendly to me . What is the problem with the default for ct ? It 's a tiny hit to cache unfriendliness ( you do n't get coherency between clients that inconsistently construct requests , or you have to redirect - and - push ) , agreed . But it gets you the agility you wanted . Oh , forgot to point out that this tends to be self - correcting . There are usually only a few different configurations of clients , so while those that support two request formats wo n't necessarily share cache hits with those that only support one , those two populations will probably happily share a cache . So to be clear , you 'd allow requests like any of the following : Personally I like this - clear and neat . Maybe I 'd change the parameter to ( so the two options are and , but that 's just me . It removes the necessity to specify the type separately and it 's ' neat ' . Plus extensible for future formats . Yeah . Pick whatever colour you like for the shed , I 'm not planning to live in it . describes a method that is more flexible again in . A better solution in my opinion ( if you set the second half of that issue aside for the moment ) . I think the color of the shed is fine for this document , and if discovery would like to do some fancy templating to enable rainbows that 's fine for that separate effort . could be much shorter . Bytes in requests matter and these are going to be used in critical conditions ( early in the connection where the number of requests you can make depend on how big they are ) . These wo n't compress well enough with HPACK to get any real benefit . A short name would be an improvement . A default value for would be even better . Could go the alt - svc approach ( ) and use two - letter combos , e.g. "" ct "" for , but really there out to be a standard for that approach ... I like the change to ' ct ' for Content - Type , but I think ' body ' is important enough where reducing it by only two bytes to ' rb ' is n't worth the change . Perhaps there could be an abbreviated ' ct ' value - e.g. ' ct = du ' ? Or , if ' ct ' is not specified , then ' application / dns - udpwireformat ' iMUST be assumed by the DNS server ?" fb31cda5950400c93017fc49096d1f7e4cf183e0 fb31cda5950400c93017fc49096d1f7e4cf183e0 draft-ietf-doh-dns-over-https-latest 804683cd1896a3f22e9aed540302b920f4e0acb5185ae135c58d1d1276579335 ['804683cd1896a3f22e9aed540302b920f4e0acb5185ae135c58d1d1276579335', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'b8ec7d5063ce550170279530749a83e23908984958cd794f9589041bb8c7d285', 'd2d81b62aa35c9d05340f4e311651515e578d7855f07f7839e34923422e284dd', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] https://tools.ietf.org/html/draft-ietf-httpbis-alt-svc-14#section-3.1) A "when using the get method, the uri path must contain a query parameter of the form content-type=ttt and another of the form body=bbbb, where ""ttt"" is the media type of the format used for the body parameter, and ""bbb"" is the content of the body encoded with base64url rfc4648. using the get method is friendlier to many http cache implementations." 19 {96, 22, 23, 58, 91} {58: 0, 23: 1, 96: 2, 22: 2, 91: 3} {19: 0, 43: 1, 64: 2, 60: 2, 80: 3} "When using the POST method the DNS query is included as the message body of the HTTP request and the Content-Type request header indicates the media type of the message. POST-ed requests are smaller than their GET equivalents." "closes issue I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a ct name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ct ( what if you mis - spell it ? ) Actually what was the reasoning behind passing Content - Type as a query parameter instead of as the Content - Type header ? Presumably any DNS API server must be able to parse headers ... Content - Type as a header refers to the message payload which is n't what we 're trying to describe here .. ( its used on POST , where it matches .. ) referring back to pull , you wrote this : "" I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ( what if you mis - spell it ? ) "" I 'm fine with having now named , but I 'm concerned by the use of a the query parameter without a value at all ... Whilst this is probably too unwieldy : /dns - query?ct = application / dns - udpwireformat&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB the use of alone just /dns - query?ct&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB Yes , I know it 's valid according to RFC 3986 , but I 'm wondering whether , since it 's very unlikely that a JSON format would be passed using a GET , we could treat the absence of as implying ? Alternatively we could allow a couple of special values for the parameter : - implies that the format is - implies that the format is There 's a potential different approach here , use a specific parameter name for each type of query . That is , has a value that is json= ... application / dns - query+json ` and so on . That way you can have multiple formats on the request ( for backwards compatibility ) . The only downside is that the backwardly compatible request will cause a hit on cache coherency , so the server might want to redirect to a canonical URL if it is concerned that the client wo n't consistently include both formats . ( That redirect can be followed by a push if you are concerned about round trips . ) This feels cache - unfriendly to me . What is the problem with the default for ct ? It 's a tiny hit to cache unfriendliness ( you do n't get coherency between clients that inconsistently construct requests , or you have to redirect - and - push ) , agreed . But it gets you the agility you wanted . Oh , forgot to point out that this tends to be self - correcting . There are usually only a few different configurations of clients , so while those that support two request formats wo n't necessarily share cache hits with those that only support one , those two populations will probably happily share a cache . So to be clear , you 'd allow requests like any of the following : Personally I like this - clear and neat . Maybe I 'd change the parameter to ( so the two options are and , but that 's just me . It removes the necessity to specify the type separately and it 's ' neat ' . Plus extensible for future formats . Yeah . Pick whatever colour you like for the shed , I 'm not planning to live in it . describes a method that is more flexible again in . A better solution in my opinion ( if you set the second half of that issue aside for the moment ) . I think the color of the shed is fine for this document , and if discovery would like to do some fancy templating to enable rainbows that 's fine for that separate effort . could be much shorter . Bytes in requests matter and these are going to be used in critical conditions ( early in the connection where the number of requests you can make depend on how big they are ) . These wo n't compress well enough with HPACK to get any real benefit . A short name would be an improvement . A default value for would be even better . Could go the alt - svc approach ( ) and use two - letter combos , e.g. "" ct "" for , but really there out to be a standard for that approach ... I like the change to ' ct ' for Content - Type , but I think ' body ' is important enough where reducing it by only two bytes to ' rb ' is n't worth the change . Perhaps there could be an abbreviated ' ct ' value - e.g. ' ct = du ' ? Or , if ' ct ' is not specified , then ' application / dns - udpwireformat ' iMUST be assumed by the DNS server ?" fb31cda5950400c93017fc49096d1f7e4cf183e0 fb31cda5950400c93017fc49096d1f7e4cf183e0 draft-ietf-doh-dns-over-https-latest d2d81b62aa35c9d05340f4e311651515e578d7855f07f7839e34923422e284dd ['804683cd1896a3f22e9aed540302b920f4e0acb5185ae135c58d1d1276579335', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'b8ec7d5063ce550170279530749a83e23908984958cd794f9589041bb8c7d285', 'd2d81b62aa35c9d05340f4e311651515e578d7855f07f7839e34923422e284dd', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] https://tools.ietf.org/html/draft-ietf-httpbis-alt-svc-14#section-3.1) B "when using the post method the dns query is included as the message body of the http request and the content-type request header indicates the media type of the message. post-ed requests are smaller than their get equivalents." 19 {96, 22, 23, 58, 91} {58: 0, 23: 1, 96: 2, 22: 2, 91: 3} {19: 0, 43: 1, 64: 2, 60: 2, 80: 3} "When using the GET method the URI path MUST contain a query parameter with the name of ct and a value indicating the media-format used for the body parameter. The value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. An empty value indicates the default application/dns-udpwireformat type (e.g. ct&body=...)." "closes issue I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a ct name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ct ( what if you mis - spell it ? ) Actually what was the reasoning behind passing Content - Type as a query parameter instead of as the Content - Type header ? Presumably any DNS API server must be able to parse headers ... Content - Type as a header refers to the message payload which is n't what we 're trying to describe here .. ( its used on POST , where it matches .. ) referring back to pull , you wrote this : "" I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ( what if you mis - spell it ? ) "" I 'm fine with having now named , but I 'm concerned by the use of a the query parameter without a value at all ... Whilst this is probably too unwieldy : /dns - query?ct = application / dns - udpwireformat&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB the use of alone just /dns - query?ct&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB Yes , I know it 's valid according to RFC 3986 , but I 'm wondering whether , since it 's very unlikely that a JSON format would be passed using a GET , we could treat the absence of as implying ? Alternatively we could allow a couple of special values for the parameter : - implies that the format is - implies that the format is There 's a potential different approach here , use a specific parameter name for each type of query . That is , has a value that is json= ... application / dns - query+json ` and so on . That way you can have multiple formats on the request ( for backwards compatibility ) . The only downside is that the backwardly compatible request will cause a hit on cache coherency , so the server might want to redirect to a canonical URL if it is concerned that the client wo n't consistently include both formats . ( That redirect can be followed by a push if you are concerned about round trips . ) This feels cache - unfriendly to me . What is the problem with the default for ct ? It 's a tiny hit to cache unfriendliness ( you do n't get coherency between clients that inconsistently construct requests , or you have to redirect - and - push ) , agreed . But it gets you the agility you wanted . Oh , forgot to point out that this tends to be self - correcting . There are usually only a few different configurations of clients , so while those that support two request formats wo n't necessarily share cache hits with those that only support one , those two populations will probably happily share a cache . So to be clear , you 'd allow requests like any of the following : Personally I like this - clear and neat . Maybe I 'd change the parameter to ( so the two options are and , but that 's just me . It removes the necessity to specify the type separately and it 's ' neat ' . Plus extensible for future formats . Yeah . Pick whatever colour you like for the shed , I 'm not planning to live in it . describes a method that is more flexible again in . A better solution in my opinion ( if you set the second half of that issue aside for the moment ) . I think the color of the shed is fine for this document , and if discovery would like to do some fancy templating to enable rainbows that 's fine for that separate effort . could be much shorter . Bytes in requests matter and these are going to be used in critical conditions ( early in the connection where the number of requests you can make depend on how big they are ) . These wo n't compress well enough with HPACK to get any real benefit . A short name would be an improvement . A default value for would be even better . Could go the alt - svc approach ( ) and use two - letter combos , e.g. "" ct "" for , but really there out to be a standard for that approach ... I like the change to ' ct ' for Content - Type , but I think ' body ' is important enough where reducing it by only two bytes to ' rb ' is n't worth the change . Perhaps there could be an abbreviated ' ct ' value - e.g. ' ct = du ' ? Or , if ' ct ' is not specified , then ' application / dns - udpwireformat ' iMUST be assumed by the DNS server ?" fb31cda5950400c93017fc49096d1f7e4cf183e0 fb31cda5950400c93017fc49096d1f7e4cf183e0 draft-ietf-doh-dns-over-https-latest 76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b ['804683cd1896a3f22e9aed540302b920f4e0acb5185ae135c58d1d1276579335', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'b8ec7d5063ce550170279530749a83e23908984958cd794f9589041bb8c7d285', 'd2d81b62aa35c9d05340f4e311651515e578d7855f07f7839e34923422e284dd', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] https://tools.ietf.org/html/draft-ietf-httpbis-alt-svc-14#section-3.1) B "when using the get method the uri path must contain a query parameter with the name of ct and a value indicating the media-format used for the body parameter. the value may either be an explicit media type (e.g. ct=application/dns-udpwireformat&body=...) or it may be empty. an empty value indicates the default application/dns-udpwireformat type (e.g. ct&body=...)." 19 {96, 22, 23, 58, 91} {58: 0, 23: 1, 96: 2, 22: 2, 91: 3} {19: 0, 43: 1, 64: 2, 60: 2, 80: 3} "When using the GET method the URI path MUST contain a query parameter with the name of body. The value of the parameter is the content of the request encoded with base64url RFC4648. Using the GET method is friendlier to many HTTP cache implementations." "closes issue I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a ct name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ct ( what if you mis - spell it ? ) Actually what was the reasoning behind passing Content - Type as a query parameter instead of as the Content - Type header ? Presumably any DNS API server must be able to parse headers ... Content - Type as a header refers to the message payload which is n't what we 're trying to describe here .. ( its used on POST , where it matches .. ) referring back to pull , you wrote this : "" I could go either way on body . You 're right that the actual mime type is a bit of a problem .. I 'll try it with a name argument without a value having the default meaning . I think that 's explicit enough , while I do n't really like totally implicit things like the absence of ( what if you mis - spell it ? ) "" I 'm fine with having now named , but I 'm concerned by the use of a the query parameter without a value at all ... Whilst this is probably too unwieldy : /dns - query?ct = application / dns - udpwireformat&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB the use of alone just /dns - query?ct&dns = AAABAAABAAAAAAAAA3d3dwdleGFtcGxlA2NvbQAAAQAB Yes , I know it 's valid according to RFC 3986 , but I 'm wondering whether , since it 's very unlikely that a JSON format would be passed using a GET , we could treat the absence of as implying ? Alternatively we could allow a couple of special values for the parameter : - implies that the format is - implies that the format is There 's a potential different approach here , use a specific parameter name for each type of query . That is , has a value that is json= ... application / dns - query+json ` and so on . That way you can have multiple formats on the request ( for backwards compatibility ) . The only downside is that the backwardly compatible request will cause a hit on cache coherency , so the server might want to redirect to a canonical URL if it is concerned that the client wo n't consistently include both formats . ( That redirect can be followed by a push if you are concerned about round trips . ) This feels cache - unfriendly to me . What is the problem with the default for ct ? It 's a tiny hit to cache unfriendliness ( you do n't get coherency between clients that inconsistently construct requests , or you have to redirect - and - push ) , agreed . But it gets you the agility you wanted . Oh , forgot to point out that this tends to be self - correcting . There are usually only a few different configurations of clients , so while those that support two request formats wo n't necessarily share cache hits with those that only support one , those two populations will probably happily share a cache . So to be clear , you 'd allow requests like any of the following : Personally I like this - clear and neat . Maybe I 'd change the parameter to ( so the two options are and , but that 's just me . It removes the necessity to specify the type separately and it 's ' neat ' . Plus extensible for future formats . Yeah . Pick whatever colour you like for the shed , I 'm not planning to live in it . describes a method that is more flexible again in . A better solution in my opinion ( if you set the second half of that issue aside for the moment ) . I think the color of the shed is fine for this document , and if discovery would like to do some fancy templating to enable rainbows that 's fine for that separate effort . could be much shorter . Bytes in requests matter and these are going to be used in critical conditions ( early in the connection where the number of requests you can make depend on how big they are ) . These wo n't compress well enough with HPACK to get any real benefit . A short name would be an improvement . A default value for would be even better . Could go the alt - svc approach ( ) and use two - letter combos , e.g. "" ct "" for , but really there out to be a standard for that approach ... I like the change to ' ct ' for Content - Type , but I think ' body ' is important enough where reducing it by only two bytes to ' rb ' is n't worth the change . Perhaps there could be an abbreviated ' ct ' value - e.g. ' ct = du ' ? Or , if ' ct ' is not specified , then ' application / dns - udpwireformat ' iMUST be assumed by the DNS server ?" fb31cda5950400c93017fc49096d1f7e4cf183e0 fb31cda5950400c93017fc49096d1f7e4cf183e0 draft-ietf-doh-dns-over-https-latest ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934 ['804683cd1896a3f22e9aed540302b920f4e0acb5185ae135c58d1d1276579335', 'ea7a65287dbf8a9e8dc41e1910bc69266f4a82bfb0b52af62e991d1dbc6f0934', 'b8ec7d5063ce550170279530749a83e23908984958cd794f9589041bb8c7d285', 'd2d81b62aa35c9d05340f4e311651515e578d7855f07f7839e34923422e284dd', '76da788e32b13779a96940a346fafc7537a62a0f6200c91da89ddc706a1aae4b'] https://tools.ietf.org/html/draft-ietf-httpbis-alt-svc-14#section-3.1) B "when using the get method the uri path must contain a query parameter with the name of body. the value of the parameter is the content of the request encoded with base64url rfc4648. using the get method is friendlier to many http cache implementations." 10 {} {} {} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as application/dns-udpwireformat, should use a DNS ID of 0 in every DNS request. HTTP semantics correlate the request and response, thus eliminating the need for the ID in a media type such as application/dns-udpwireformat." "And explain why more clearly . this is editorial - should me merged imo I fully disagree that making something a RFC 2119 requirement is "" editorial "" . In this case , it could have surprising effects on the DNS server side of a DOH server . It 's worth discussing , but not just sweeping it in with "" editorial "" . when I wrote that text I meant it to be SHOULD - its a typo ; that 's why its editorial :) If its not normative then it really has no meaning in that paragraph ." c6c6ee1fdbab4c7d72de2c135daa6d18c5072309 c6c6ee1fdbab4c7d72de2c135daa6d18c5072309 draft-ietf-doh-dns-over-https-latest 62668e5503e44cc85ef8ce3f3c8ed779c3af560b441f57933f7dc574aceb54b1 ['f709523fb9e4d455214ae45763bea2e9c68d3325f39b0ab7b02d9b22efd51663', '62668e5503e44cc85ef8ce3f3c8ed779c3af560b441f57933f7dc574aceb54b1'] A "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as application/dns-udpwireformat, should use a dns id of 0 in every dns request. http semantics correlate the request and response, thus eliminating the need for the id in a media type such as application/dns-udpwireformat." 10 {} {} {} "In order to maximize cache friendliness, DNS API clients using media formats that include DNS ID, such as application/dns-udpwireformat, SHOULD use a DNS ID of 0 in every DNS request. HTTP correlates request and response, thus eliminating the need for the ID in a media type such as application/dns-udpwireformat and the use of a varying DNS ID can cause semantically equivalent DNS queries to be cached separately." "And explain why more clearly . this is editorial - should me merged imo I fully disagree that making something a RFC 2119 requirement is "" editorial "" . In this case , it could have surprising effects on the DNS server side of a DOH server . It 's worth discussing , but not just sweeping it in with "" editorial "" . when I wrote that text I meant it to be SHOULD - its a typo ; that 's why its editorial :) If its not normative then it really has no meaning in that paragraph ." c6c6ee1fdbab4c7d72de2c135daa6d18c5072309 c6c6ee1fdbab4c7d72de2c135daa6d18c5072309 draft-ietf-doh-dns-over-https-latest f709523fb9e4d455214ae45763bea2e9c68d3325f39b0ab7b02d9b22efd51663 ['f709523fb9e4d455214ae45763bea2e9c68d3325f39b0ab7b02d9b22efd51663', '62668e5503e44cc85ef8ce3f3c8ed779c3af560b441f57933f7dc574aceb54b1'] B "in order to maximize cache friendliness, dns api clients using media formats that include dns id, such as application/dns-udpwireformat, should use a dns id of 0 in every dns request. http correlates request and response, thus eliminating the need for the id in a media type such as application/dns-udpwireformat and the use of a varying dns id can cause semantically equivalent dns queries to be cached separately." 6 {65, 67, 36, 73, 10, 75, 113} {10: 0, 65: 1, 73: 2, 75: 2, 67: 3, 36: 3, 113: 3} {6: 0, 61: 1, 131: 2, 33: 2, 133: 3, 128: 3, 199: 3} "The path SHOULD be ""/.well-known/dns-query"" but a different path can be used if the DNS API Client has prior knowledge about a DNS API service on a different path at the origin being used. (See iana for the registration of this in the well-known URI registry.) Using the well-known path allows automated discovery of a DNS API Service, and also helps contextualize DNS Query requests pushed over an active HTTP/2 connection." "Firstly , mention that the URI can be anything . Then explain why the well - known exists . The well - known URI exists for two reasons : aid in discovery , for cases where the discovery process produces only a domain name ( or domain name and port ) contextualize server pushes The first does n't need a SHOULD . The draft gets stuck into describing pieces of the URL , but neglects to point out some obvious things , like what the URL is for . It could probably use some more clarity about the configuration and discovery point as well ." 32870834d9bda90e18f88ff03c7b649aadeb7ec5 32870834d9bda90e18f88ff03c7b649aadeb7ec5 draft-ietf-doh-dns-over-https-latest be06cc82adca92bc01b6c85e0c188f2fe799ddd025cec4a8e07564756b8f6a24 ['10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', 'be06cc82adca92bc01b6c85e0c188f2fe799ddd025cec4a8e07564756b8f6a24'] A "the path should be ""/.well-known/dns-query"" but a different path can be used if the dns api client has prior knowledge about a dns api service on a different path at the origin being used. (see iana for the registration of this in the well-known uri registry.) using the well-known path allows automated discovery of a dns api service, and also helps contextualize dns query requests pushed over an active http/2 connection." 6 {65, 67, 36, 73, 10, 75, 113} {10: 0, 65: 1, 73: 2, 75: 2, 67: 3, 36: 3, 113: 3} {6: 0, 61: 1, 131: 2, 33: 2, 133: 3, 128: 3, 199: 3} "To make a DNS API query, a DNS API client sends an HTTP request to the URI of the DNS API." "Firstly , mention that the URI can be anything . Then explain why the well - known exists . The well - known URI exists for two reasons : aid in discovery , for cases where the discovery process produces only a domain name ( or domain name and port ) contextualize server pushes The first does n't need a SHOULD . The draft gets stuck into describing pieces of the URL , but neglects to point out some obvious things , like what the URL is for . It could probably use some more clarity about the configuration and discovery point as well ." 32870834d9bda90e18f88ff03c7b649aadeb7ec5 32870834d9bda90e18f88ff03c7b649aadeb7ec5 draft-ietf-doh-dns-over-https-latest dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b ['10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', 'be06cc82adca92bc01b6c85e0c188f2fe799ddd025cec4a8e07564756b8f6a24'] B "to make a dns api query, a dns api client sends an http request to the uri of the dns api." 6 {65, 67, 36, 73, 10, 75, 113} {10: 0, 65: 1, 73: 2, 75: 2, 67: 3, 36: 3, 113: 3} {6: 0, 61: 1, 131: 2, 33: 2, 133: 3, 128: 3, 199: 3} "A client can be configured with a DNS API URI, or it can discover the URI. This document defines a well-known URI path of ""/.well-known/ dns-query"" so that a discovery process that produces a domain name or domain name and port can be used to construct the DNS API URI. (See iana for the registration of this in the well-known URI registry.) DNS API servers SHOULD use this well-known path to help contextualize DNS Query requests that use server push RFC7540." "Firstly , mention that the URI can be anything . Then explain why the well - known exists . The well - known URI exists for two reasons : aid in discovery , for cases where the discovery process produces only a domain name ( or domain name and port ) contextualize server pushes The first does n't need a SHOULD . The draft gets stuck into describing pieces of the URL , but neglects to point out some obvious things , like what the URL is for . It could probably use some more clarity about the configuration and discovery point as well ." 32870834d9bda90e18f88ff03c7b649aadeb7ec5 32870834d9bda90e18f88ff03c7b649aadeb7ec5 draft-ietf-doh-dns-over-https-latest 10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d ['10e7da4d3cf1d11a91aebcd54e47cde894304acead6f52a9be33bb9a8d74416d', 'dcffaf6c182ca56cfe6e98446e5404bb7867c351831ef96925ee20ccef36977b', 'be06cc82adca92bc01b6c85e0c188f2fe799ddd025cec4a8e07564756b8f6a24'] B "a client can be configured with a dns api uri, or it can discover the uri. this document defines a well-known uri path of ""/.well-known/ dns-query"" so that a discovery process that produces a domain name or domain name and port can be used to construct the dns api uri. (see iana for the registration of this in the well-known uri registry.) dns api servers should use this well-known path to help contextualize dns query requests that use server push rfc7540." 4 {13, 14} {13: 0, 14: 1} {4: 0, 1: 1} "The primary use case is to prevent on-path network devices from interfering with native DNS operations. This interference includes, but is not limited to, spoofing DNS responses, blocking DNS requests, and tracking. HTTP authentication and proxy friendliness are expected to make this protocol function in some environments where DNS directly on TLS (RFC7858) would not." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." acd907f74a71efa42a2645633c02e6b63ecd6700 acd907f74a71efa42a2645633c02e6b63ecd6700 draft-ietf-doh-dns-over-https-latest 173282eb99c5732cbc6f37047494b486e8baf820c72abb432a9ffcc8c9eac357 ['173282eb99c5732cbc6f37047494b486e8baf820c72abb432a9ffcc8c9eac357', '218e8c30eba961a3f323f724d0915eefaa0dd547edf385e8d9d1fdaec2c323eb'] A "the primary use case is to prevent on-path network devices from interfering with native dns operations. this interference includes, but is not limited to, spoofing dns responses, blocking dns requests, and tracking. http authentication and proxy friendliness are expected to make this protocol function in some environments where dns directly on tls (rfc7858) would not." 4 {13, 14} {13: 0, 14: 1} {4: 0, 1: 1} "The primary use case is to prevent on-path network devices from interfering with native DNS operations. This interference includes, but is not limited to, spoofing DNS responses, blocking DNS requests, and tracking. HTTP authentication and proxy friendliness are expected to make this protocol function in some environments where unsecured DNS (DNS) or DNS directly on TLS (RFC7858) would not." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." acd907f74a71efa42a2645633c02e6b63ecd6700 acd907f74a71efa42a2645633c02e6b63ecd6700 draft-ietf-doh-dns-over-https-latest 218e8c30eba961a3f323f724d0915eefaa0dd547edf385e8d9d1fdaec2c323eb ['173282eb99c5732cbc6f37047494b486e8baf820c72abb432a9ffcc8c9eac357', '218e8c30eba961a3f323f724d0915eefaa0dd547edf385e8d9d1fdaec2c323eb'] B "the primary use case is to prevent on-path network devices from interfering with native dns operations. this interference includes, but is not limited to, spoofing dns responses, blocking dns requests, and tracking. http authentication and proxy friendliness are expected to make this protocol function in some environments where unsecured dns (dns) or dns directly on tls (rfc7858) would not." 1 {13, 14} {14: 0, 13: 1} {1: 0, 4: 1} "The primary one is to prevent on-path network devices from interfering with native DNS operations. This interference includes, but is not limited to, spoofing DNS responses, blocking DNS requests, and tracking. HTTP authentication and proxy friendliness are expected to make this protocol function in some environments where DNS directly on TLS (RFC7858) would not." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 7f70a81ae2ff3805d16631b87d239358e7f0ae3ecacaddeded83deabc600536e ['173282eb99c5732cbc6f37047494b486e8baf820c72abb432a9ffcc8c9eac357', '7f70a81ae2ff3805d16631b87d239358e7f0ae3ecacaddeded83deabc600536e'] A "the primary one is to prevent on-path network devices from interfering with native dns operations. this interference includes, but is not limited to, spoofing dns responses, blocking dns requests, and tracking. http authentication and proxy friendliness are expected to make this protocol function in some environments where dns directly on tls (rfc7858) would not." 1 {13, 14} {14: 0, 13: 1} {1: 0, 4: 1} "The primary use case is to prevent on-path network devices from interfering with native DNS operations. This interference includes, but is not limited to, spoofing DNS responses, blocking DNS requests, and tracking. HTTP authentication and proxy friendliness are expected to make this protocol function in some environments where DNS directly on TLS (RFC7858) would not." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 173282eb99c5732cbc6f37047494b486e8baf820c72abb432a9ffcc8c9eac357 ['173282eb99c5732cbc6f37047494b486e8baf820c72abb432a9ffcc8c9eac357', '7f70a81ae2ff3805d16631b87d239358e7f0ae3ecacaddeded83deabc600536e'] B "the primary use case is to prevent on-path network devices from interfering with native dns operations. this interference includes, but is not limited to, spoofing dns responses, blocking dns requests, and tracking. http authentication and proxy friendliness are expected to make this protocol function in some environments where dns directly on tls (rfc7858) would not." 1 {} {} {} "A secondary use case is web applications that want to access DNS information. Standardizing an HTTPS mechanism allows this to be done in a way consistent with the cross-origin resource sharing CORS security model of the web and also integrate the caching mechanisms of DNS with those of HTTP. These applications may be interested in using a different media type than traditional clients." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 9c94625b37b6e7442b593c48aaa06dd5c278735740e3e9d2bde124e53f098925 ['a7b7468722fb4d6af80c0be88e61a7b35ca8ecb16ec08f6687973d25603d99f6', '9c94625b37b6e7442b593c48aaa06dd5c278735740e3e9d2bde124e53f098925'] A "a secondary use case is web applications that want to access dns information. standardizing an https mechanism allows this to be done in a way consistent with the cross-origin resource sharing cors security model of the web and also integrate the caching mechanisms of dns with those of http. these applications may be interested in using a different media type than traditional clients." 1 {} {} {} "A secondary use case is web applications that want to access DNS information. Standardizing an HTTPS mechanism allows this to be done in a way consistent with the cross-origin resource sharing security model of the web CORS and also integrate the caching mechanisms of DNS with those of HTTP. These applications may be interested in using a different media type than traditional clients." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest a7b7468722fb4d6af80c0be88e61a7b35ca8ecb16ec08f6687973d25603d99f6 ['a7b7468722fb4d6af80c0be88e61a7b35ca8ecb16ec08f6687973d25603d99f6', '9c94625b37b6e7442b593c48aaa06dd5c278735740e3e9d2bde124e53f098925'] B "a secondary use case is web applications that want to access dns information. standardizing an https mechanism allows this to be done in a way consistent with the cross-origin resource sharing security model of the web cors and also integrate the caching mechanisms of dns with those of http. these applications may be interested in using a different media type than traditional clients." 1 {92, 93} {93: 0, 92: 1} {1: 0, 43: 1} "The protocol must use a secure transport that meets the requirements for modern https://." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 93e06618edf349357925c6d2cc9e20d546c3e8725e2985947420152cb052dca4 ['c7c2b1c39835e8b1c8b90efe7f1d83b71b87e390ab1a607b555ef8d1484bff2e', '93e06618edf349357925c6d2cc9e20d546c3e8725e2985947420152cb052dca4'] A "the protocol must use a secure transport that meets the requirements for modern https://." 1 {92, 93} {93: 0, 92: 1} {1: 0, 43: 1} "The protocol must use a secure transport that meets the requirements for modern HTTPS." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest c7c2b1c39835e8b1c8b90efe7f1d83b71b87e390ab1a607b555ef8d1484bff2e ['c7c2b1c39835e8b1c8b90efe7f1d83b71b87e390ab1a607b555ef8d1484bff2e', '93e06618edf349357925c6d2cc9e20d546c3e8725e2985947420152cb052dca4'] B "the protocol must use a secure transport that meets the requirements for modern https." 1 {81, 66} {81: 0, 66: 1} {1: 0, 47: 1} "The media type is ""application/dns-udpwireformat"". The body is the DNS on-the-wire format is defined in RFC1035. The body MUST be encoded with base64url RFC4648. Padding characters for base64url MUST NOT be included." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest cba05075bfc72a282c262d19d45626641dcae7ee2dd0c52dd496104aa962fb74 ['b27f7bbd918fd7d59785f7552ddbc104f4d2ab8f1b3b1d38226d150eca5e95f3', '252bf1771c56d941c53f917f18a7a05405c78e5091b4a66e880d3ca5828c0a89', '8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43', 'cba05075bfc72a282c262d19d45626641dcae7ee2dd0c52dd496104aa962fb74'] A "the media type is ""application/dns-udpwireformat"". the body is the dns on-the-wire format is defined in rfc1035. the body must be encoded with base64url rfc4648. padding characters for base64url must not be included." 1 {81, 66} {81: 0, 66: 1} {1: 0, 47: 1} "The media type is ""application/dns-udpwireformat"". The body is the DNS on-the-wire format is defined in RFC1035." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 252bf1771c56d941c53f917f18a7a05405c78e5091b4a66e880d3ca5828c0a89 ['b27f7bbd918fd7d59785f7552ddbc104f4d2ab8f1b3b1d38226d150eca5e95f3', '252bf1771c56d941c53f917f18a7a05405c78e5091b4a66e880d3ca5828c0a89', '8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43', 'cba05075bfc72a282c262d19d45626641dcae7ee2dd0c52dd496104aa962fb74'] B "the media type is ""application/dns-udpwireformat"". the body is the dns on-the-wire format is defined in rfc1035." 1 {81, 66} {81: 0, 66: 1} {1: 0, 47: 1} "When using the GET method, the body MUST be encoded with base64url RFC4648. Padding characters for base64url MUST NOT be included." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest b27f7bbd918fd7d59785f7552ddbc104f4d2ab8f1b3b1d38226d150eca5e95f3 ['b27f7bbd918fd7d59785f7552ddbc104f4d2ab8f1b3b1d38226d150eca5e95f3', '252bf1771c56d941c53f917f18a7a05405c78e5091b4a66e880d3ca5828c0a89', '8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43', 'cba05075bfc72a282c262d19d45626641dcae7ee2dd0c52dd496104aa962fb74'] B "when using the get method, the body must be encoded with base64url rfc4648. padding characters for base64url must not be included." 1 {81, 66} {81: 0, 66: 1} {1: 0, 47: 1} When using the POST method, the body is not encoded. "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43 ['b27f7bbd918fd7d59785f7552ddbc104f4d2ab8f1b3b1d38226d150eca5e95f3', '252bf1771c56d941c53f917f18a7a05405c78e5091b4a66e880d3ca5828c0a89', '8ca45703e515b8e850c417ec388dec94a19e8bb94e5f2a01e071b367daa73f43', 'cba05075bfc72a282c262d19d45626641dcae7ee2dd0c52dd496104aa962fb74'] B when using the post method, the body is not encoded. 1 {38, 70, 19, 20, 118, 94} {20: 0, 94: 1, 19: 2, 118: 3, 70: 4, 38: 4} {1: 0, 43: 1, 78: 2, 142: 3, 144: 4, 122: 4} "At the time this is published, the response types are works in progress. The only known response type is ""application/dns- udpwireformat"", but it is likely that at least one JSON-based response format will be defined in the future." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 2b2de1b660ea8fab1ecea8d734791fb81035046714e72c96ca5690a5673df2b4 ['2b2de1b660ea8fab1ecea8d734791fb81035046714e72c96ca5690a5673df2b4', 'cac4379edb3ea1f1be5066e668a429264e4c6e1bb43cc1af2bfbcb4615557c94'] A "at the time this is published, the response types are works in progress. the only known response type is ""application/dns- udpwireformat"", but it is likely that at least one json-based response format will be defined in the future." 1 {38, 70, 19, 20, 118, 94} {20: 0, 94: 1, 19: 2, 118: 3, 70: 4, 38: 4} {1: 0, 43: 1, 78: 2, 142: 3, 144: 4, 122: 4} "At the time this is published, the response types are works in progress. The only known response type is ""application/dns- udpwireformat"", but it is possible that at least one JSON-based response format will be defined in the future." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest cac4379edb3ea1f1be5066e668a429264e4c6e1bb43cc1af2bfbcb4615557c94 ['2b2de1b660ea8fab1ecea8d734791fb81035046714e72c96ca5690a5673df2b4', 'cac4379edb3ea1f1be5066e668a429264e4c6e1bb43cc1af2bfbcb4615557c94'] B "at the time this is published, the response types are works in progress. the only known response type is ""application/dns- udpwireformat"", but it is possible that at least one json-based response format will be defined in the future." 1 {121, 59, 61, 47} {61: 0, 59: 1, 121: 2, 47: 2} {1: 0, 34: 1, 165: 2, 45: 2} "In order to satisfy the security requirements of DNS over HTTPS, this protocol MUST use HTTP/2 RFC7540 or its successors. HTTP/2 enforces a modern TLS profile necessary for achieving the security requirements of this protocol." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 874fb79184f0a2251a5db70d2f620d59d4de93e394907fa8951acd444bae2df7 ['874fb79184f0a2251a5db70d2f620d59d4de93e394907fa8951acd444bae2df7', '826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8', 'c1f5653eaf6f3b68aca904e1576a30b8fbd2609d4b8965f67f24de0dfdc79af3'] A "in order to satisfy the security requirements of dns over https, this protocol must use http/2 rfc7540 or its successors. http/2 enforces a modern tls profile necessary for achieving the security requirements of this protocol." 1 {121, 59, 61, 47} {61: 0, 59: 1, 121: 2, 47: 2} {1: 0, 34: 1, 165: 2, 45: 2} "The messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, priority, parallelism, and header compression. For this additional reason, this protocol MUST use HTTP/2 RFC7540 or its successors." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest c1f5653eaf6f3b68aca904e1576a30b8fbd2609d4b8965f67f24de0dfdc79af3 ['874fb79184f0a2251a5db70d2f620d59d4de93e394907fa8951acd444bae2df7', '826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8', 'c1f5653eaf6f3b68aca904e1576a30b8fbd2609d4b8965f67f24de0dfdc79af3'] A "the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, priority, parallelism, and header compression. for this additional reason, this protocol must use http/2 rfc7540 or its successors." 1 {121, 59, 61, 47} {61: 0, 59: 1, 121: 2, 47: 2} {1: 0, 34: 1, 165: 2, 45: 2} "This protocol MUST use HTTP/2 RFC7540 or its successors in order to satisfy the security requirements of DNS over HTTPS. Further, the messages in classic UDP based DNS RFC1035 are inherently unordered and have low overhead. A competitive HTTP transport needs to support reordering, priority, parallelism, and header compression, all of which are supported by HTTP/2 RFC7540 or its successors." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8 ['874fb79184f0a2251a5db70d2f620d59d4de93e394907fa8951acd444bae2df7', '826e2dd815b26e4cd084342bb806335bb2998c5cef73c20d8dfd9eaa98619db8', 'c1f5653eaf6f3b68aca904e1576a30b8fbd2609d4b8965f67f24de0dfdc79af3'] B "this protocol must use http/2 rfc7540 or its successors in order to satisfy the security requirements of dns over https. further, the messages in classic udp based dns rfc1035 are inherently unordered and have low overhead. a competitive http transport needs to support reordering, priority, parallelism, and header compression, all of which are supported by http/2 rfc7540 or its successors." 1 {8, 68, 69, 7} {68: 0, 69: 1, 7: 1, 8: 2} {1: 0, 36: 1, 34: 1, 99: 2} "Running DNS over https:// relies on the security of the underlying HTTP connection. By requiring at least RFC7540 levels of support for TLS this protocol expects to use current best practices for secure transport." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 28ab9b70dd2b5e655cca338f994efeb97f471ba9607956c7220fe11401f5aeef ['8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2', '28ab9b70dd2b5e655cca338f994efeb97f471ba9607956c7220fe11401f5aeef'] A "running dns over https:// relies on the security of the underlying http connection. by requiring at least rfc7540 levels of support for tls this protocol expects to use current best practices for secure transport." 1 {8, 68, 69, 7} {68: 0, 69: 1, 7: 1, 8: 2} {1: 0, 36: 1, 34: 1, 99: 2} "Running DNS over HTTPS relies on the security of the underlying HTTP connection. By requiring at least RFC7540 levels of support for TLS, this protocol expects to use current best practices for secure transport." "I think that we should specify behaviour for truncated responses with the TC bit more clearly . Right now the text is a little vague . There seem to be three options : the generation of an HTTP error the forwarding of the truncated response allow either the use of truncated responses and recommend fallback to TCP if the DNS API server is just making DNS requests to another server I think that the draft current vacilates in the direction of 3 , but I tend to like 1 or 4 . Choosing 4 would n't be consistent with the dnsop DNS - over - HTTP draft using DOH though . is the right answer , so the text should just be stronger about it . If DOH requires or , that means that the DNS server code will need to be heavily reworked to capture these from going out , which seems pointless . DOH clients can act like normal stubs : either they do or do n't know about truncation and . I agree - I do n't think doh needs to behave differently here than any other datagram based client - i.e. its not a transport error . can you do a pr to clarify ? Will do ." a767ba8163effb016ef64fad629e3d43cd0c504d a767ba8163effb016ef64fad629e3d43cd0c504d draft-ietf-doh-dns-over-https-latest 8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2 ['8f965ea040aacecbfe3c7504d2021c9001aedf7006170fe4c0b6bff181ad09f2', '28ab9b70dd2b5e655cca338f994efeb97f471ba9607956c7220fe11401f5aeef'] B "running dns over https relies on the security of the underlying http connection. by requiring at least rfc7540 levels of support for tls, this protocol expects to use current best practices for secure transport."