text stringlengths 0 1.49k |
|---|
**Brian Ketelsen:** \[laughs\] It's good! |
**Filippo Valsorda:** ...both you, and you know who you are, everyone else. |
**Erik St. Martin:** Yeah, it was me, Brian, Dave Cheney... Who else? Yeah, so... Living proof here - if you wanna rehearse early, we are always welcome to have people in our hotel rooms and rehearse. |
**Brian Ketelsen:** And by early, that means 2 AM the night before you're gonna talk. \[laughter\] It's part of the burden of being a GopherCon organizer. |
**Erik St. Martin:** So I'll make two YouTube recommendations then. There's the one I've just mistakenly confused - Goerge Tankersley did the talk about Crypto, but Filippo's was really good too, about cgo. |
**Filippo Valsorda:** Yeah, the black magic, cgo and how you definitely shouldn't use it. But if you really, really have to, well, then this is how you can make it tolerable. That was the punchline. |
**Erik St. Martin:** How to make cgo tolerable. \[laughter\] |
**Carlisia Thompson:** But Filippo does have a talk on TLS 1.3... Where did you give that talk? |
**Filippo Valsorda:** That would be 33c3, the Chaos Computing Club conference in Hamburg. You can find it if you search for 33c3 TLS 1.3. There we go through all the Crypto parts of this TLS 1.3 effort, and about the Go part - there's nothing published just yet. You can find blog posts on the Gopher Academy Advent list... |
The more Crypto part... I don't know... I mean, maybe GopherCon? This is probably bad taste, I'll shut up. |
**Erik St. Martin:** I think somebody in the GoTime FM channel just said, "I think that everybody crashed your server." \[laughs\] |
**Filippo Valsorda:** Oh boy, did you?! No, I think it's the HTTP part. |
**Brian Ketelsen:** Oh, yeah... You can't SSH to an HTTP server. |
**Erik St. Martin:** Oh, wait... There's Cory. |
**Brian Ketelsen:** Cory LaNou, trying to SSH into an HTTP server. Two different protocols there, Turbo. |
**Filippo Valsorda:** Hey, you never know... \[laughter\] |
**Brian Ketelsen:** It can't be the tweets -- oh, it's the tweets problem! Oops, Adam! |
**Erik St. Martin:** \[23:54\] That was Adam. \[laughter\] So I think now is probably a good time to take our first sponsored break. Our first sponsor today is StackImpact. |
**Break:** \[24:07\] to \[24:57\] |
**Erik St. Martin:** We are back, talking to Filippo about Crypto and TLS and all the great things he's doing at CloudFlare. What else are you working on these days? I know that you've had some interesting Caddy as well, and we've had Matt Holt on the show, too. |
**Filippo Valsorda:** Yeah, I like what Matt is doing a lot, and I planned to use Caddy for a little experiment of mine. This is completely a side project... Something that I don't know if people realize or actually care about not being huge Crypto nerds is that the Go binaries are completely reproducible. If you take ... |
Now, that's super nice because it means that you can take, for example, the Caddy build server, which is a nice server that does builds for you and gets you this single binary that you can deploy, and you can prove that Matt is not an evil spy with a plan to conquer the world - sorry, Matt! - you can reproduce the bina... |
**Brian Ketelsen:** Yeah, you lost me there.. What's CT? |
**Filippo Valsorda:** Sorry, that was too much of a tangent... CT is a way to get CAs, the ones that sign TLS certificates more accountable by forcing them to disclose all the certificates they sign to a log where - long story short - they can't be hidden or removed. You could use the same ideas to make a build server ... |
That's one step further, it's called binary transparency; it doesn't have that much to do with Go, but Go is a very good language to start this, because getting reproducible builds is incredibly hard with anything else. |
The Debian project has been primed very hard to get the whole Deb repositories reproducible, and they're jumping through hoops that you wouldn't even imagine. With Go instead, you just set the same GOPATH and you're done. |
**Erik St. Martin:** \[28:03\] Yeah, you would just have to make sure you have the same Go toolchain, right? The right Go version, because theoretically it wouldn't produce the same binary if you had a different Go version. |
**Filippo Valsorda:** Oh, yeah. |
**Erik St. Martin:** And I guess you would have to track the binary for each platform, too... Because the resulting binary would change for the Windows build versus the Linux build versus the Arm build, or... |
**Filippo Valsorda:** Yeah, you have the whole metrics, architectures and operating systems, but those are like 20, or something. |
**Brian Ketelsen:** It gets bigger every release, that's the best part. |
**Filippo Valsorda:** Yup. |
**Brian Ketelsen:** We're supporting 32-bit Spark on a Raspberry Pi now. |
**Erik St. Martin:** Is there really a 32-bit Spark? |
**Brian Ketelsen:** Not on the Raspberry Pi, no... No, I made that up, but it was apparently nowhere near as funny as it should have been. |
**Erik St. Martin:** You're confusing me here. |
**Brian Ketelsen:** I can run Go on my DEC Alpha, how's that? |
**Erik St. Martin:** \[laughs\] Break out the Commodore... |
**Brian Ketelsen:** I remember back in the days when a DEC Alpha was a big deal. |
**Filippo Valsorda:** I mean, not even joking, you can run Go on mainframes now. |
**Erik St. Martin:** Yeah... It's insane watching Go take off... It blows my mind, in just a couple of years, everywhere that we see Go. So you mentioned something in our show document about latency profiling, and Camlistore too. |
**Filippo Valsorda:** Yeah, let's see... Let's go for spoilers first - latency profiling is what I plan to talk about at GopherCon India; essentially, the story there is that we are used to all the profiling tools, and they're super nice, very easy to use, and you can figure out what functions take the most CPU in your... |
It's rare that APIs are truthful about how many things concurrently they can process; it's absolutely possible most of the times - this is debatable - but surely there's a huge interest in reducing latency, making APIs return as quickly as possible. So Go does provide the tools to inspect what functions are slow are ju... |
**Erik St. Martin:** That's actually really interesting, because that debate comes up a lot, especially when talking about garbage collection, that difference between latency and throughput. Both of them kind of represent speed, right? And deciding which one you wanna optimize for is difficult. |
**Filippo Valsorda:** That's a very good point. In Go, we almost explicitly stated that we optimize for latency, at least in the garbage collector, by making it faster and faster in terms of pauses, but slightly slower in terms of CPU and throughput. So it's interesting that the profiling tools haven't caught up to the... |
**Erik St. Martin:** \[31:59\] Yeah, that's true. So you're talking about this at GopherCon India? |
**Filippo Valsorda:** Yeah, that's the plan. |
**Erik St. Martin:** Which is coming up in February... I'm trying to remember the exact dates. Does anybody remember them off-hand? |
**Brian Ketelsen:** I don't. |
**Filippo Valsorda:** February 21st? I mean, don't book tickets based on my recollection, but... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.