text
stringlengths
0
1.49k
[1274.10 --> 1274.54] Correct.
[1274.54 --> 1279.44] So you did the Vigo talk.
[1279.94 --> 1282.12] But yeah, we were all hanging out with doing it.
[1282.24 --> 1284.02] Yeah, Filippo wrote the blog post.
[1284.42 --> 1284.76] Yes.
[1285.30 --> 1289.38] But by the way, I have to say, I'm extremely grateful for that talk rehearsal.
[1290.72 --> 1291.52] That's good.
[1292.28 --> 1295.62] But both you and, you know, you know who you are for everyone else.
[1296.10 --> 1301.28] Yeah, it was me, Brian, Dave Cheney, who else?
[1301.28 --> 1301.68] Yes.
[1302.24 --> 1302.54] But yeah.
[1302.94 --> 1303.10] Yeah.
[1303.18 --> 1304.78] So living proof here.
[1305.04 --> 1305.20] Yeah.
[1305.32 --> 1311.00] If you want to rehearse early, we are always welcome to have people in our hotel rooms and rehearse.
[1311.62 --> 1314.46] And by early, that means 2 a.m. the night before you're going to talk.
[1317.16 --> 1320.62] That's part of the burden of being a GopherCon organizer.
[1320.62 --> 1325.06] So I'll make two YouTube recommendations then.
[1325.38 --> 1327.80] There's the one I'm just mistakenly confused.
[1328.02 --> 1332.38] George Tankersley did the talk about crypto, but Filippo is just really good, too, about Sego.
[1332.90 --> 1333.36] Yeah.
[1333.54 --> 1337.76] The black magic of Sego and how you definitely shouldn't use it.
[1337.84 --> 1342.78] But if you really, really, really have to, well, then this is how you can make it tolerable.
[1342.78 --> 1346.18] That was the punchline.
[1347.38 --> 1349.00] How to make Sego tolerable.
[1351.04 --> 1354.56] Filippo does have a talk on TLS 1.3.
[1355.10 --> 1356.56] Where did you give that talk?
[1357.18 --> 1363.16] That would be 33C3, the Coast Computing Club conference in Hamburg.
[1363.16 --> 1368.36] And, yeah, you can find it if you search for 33C3 TLS 1.3.
[1368.88 --> 1373.72] There we go through all the crypto parts of this TLS 1.3 effort.
[1374.38 --> 1379.10] And about the Go part, there's nothing published just yet.
[1379.64 --> 1384.52] You can find a blog post on the Gopher Academy Advent list,
[1384.94 --> 1389.52] which is a bunch of lessons learned from exposing Go server to the Internet,
[1389.52 --> 1394.38] because that's effectively what we did with the Go reverse proxy.
[1395.06 --> 1399.94] And the more crypto part, I don't know, I mean, maybe Gopher call?
[1400.60 --> 1402.08] This is probably in bad taste.
[1402.32 --> 1403.00] I'll shut up.
[1404.98 --> 1407.66] I think somebody in the Gotime FM channel just said,
[1407.76 --> 1409.68] I think everybody crashed your server.
[1410.72 --> 1412.34] Oh, boy, did you?
[1413.04 --> 1413.52] Oh, no.
[1414.04 --> 1416.84] No, no, I think it's the HTTP part.
[1417.44 --> 1418.74] Oh, yeah.
[1418.74 --> 1418.88] Yeah.
[1419.52 --> 1421.32] You can't SSH to an HTTP server.
[1421.70 --> 1422.10] Oh, wait.
[1422.46 --> 1422.86] Corey.
[1423.68 --> 1426.54] Corey Lanoue trying to SSH into an HTTP server.
[1426.98 --> 1428.48] Two different protocols there, Turbo.
[1429.56 --> 1430.66] Hey, you never know.
[1432.96 --> 1433.84] Copy the tweet.
[1433.94 --> 1435.18] Oh, it's the tweets problem.
[1435.86 --> 1436.14] Oops.
[1436.58 --> 1436.88] Adam.
[1437.70 --> 1438.62] Oh, that was Adam.
[1441.16 --> 1446.20] So I think now is probably a good time to take our first sponsor break.
[1446.20 --> 1449.20] And our first sponsor today is Stack Impact.
[1449.52 --> 1455.74] When it comes to profiling and monitoring the performance of your Go applications,
[1456.20 --> 1460.44] Stack Impact is a great service to help you and your team laser focus on hotspot profiling,
[1460.68 --> 1462.94] bottleneck tracing, health monitoring, and more.
[1463.32 --> 1467.04] Stack Impact gives you the necessary historical deep dive performance visibility
[1467.04 --> 1471.40] into your Go application's execution so you can discover and resolve performance bottlenecks
[1471.40 --> 1472.90] with line of code precision.
[1473.30 --> 1478.32] Technically, Stack Impact makes Go's built-in profiling capabilities usable in a production
[1478.32 --> 1478.84] environment.
[1479.18 --> 1480.84] Stack Impact does everything automatically.
[1480.98 --> 1484.12] There's no need to run commands or waste time specifying what to monitor.
[1484.36 --> 1487.50] They've even put their Go agent on GitHub under the BSD license.
[1487.50 --> 1491.64] So if you need to focus on the performance of your Go applications, check out Stack Impact.
[1492.06 --> 1496.60] Head over to stackimpact.com slash GoTime to learn more and tell them Brian from GoTime
[1496.60 --> 1497.08] sent you.
[1497.08 --> 1508.40] We are back talking to Filippo about crypto and TLS and all the great things he's doing
[1508.40 --> 1509.46] at Cloudflare.
[1510.18 --> 1512.94] So what else are you working on these days?
[1512.94 --> 1517.06] I know that you've had some interest in Caddy as well, and we've had Matt Holt on the show
[1517.06 --> 1517.42] too.
[1517.42 --> 1518.78] Oh, yeah, yeah.
[1518.88 --> 1520.90] I mean, I like what Matt is doing a lot.
[1521.16 --> 1525.26] And I planned to use Caddy for a little experiment of mine.
[1525.80 --> 1528.14] Now, this is a complete, complete aside project.
[1528.80 --> 1534.24] Something that I don't know if people realize or actually care about not being huge crypto
[1534.24 --> 1537.98] nerds is that the Go binaries are completely reproducible.
[1538.16 --> 1545.36] So if you take the same Go path, the same Go root, the same Go compiler, and the same code
[1545.36 --> 1552.94] base, of course, and compile it on an OpenBSD machine and a completely different Linux machine
[1552.94 --> 1559.44] and cross-compile them to the same target, the resulting binaries are identical, like
[1559.44 --> 1560.54] byte by byte.
[1561.00 --> 1563.36] And they will be forever, whoever builds that.
[1563.90 --> 1568.88] Now, that's super nice because it means that you can take, for example, the Caddy build
[1568.88 --> 1573.40] server, which is a nice server that does builds for you and gets you to this single binary
[1573.40 --> 1574.32] that you can deploy.
[1574.76 --> 1580.58] And you can prove that Matt is not an evil spy with a plan to conquer the world.
[1580.94 --> 1581.36] Sorry, Matt.
[1582.08 --> 1588.36] And you can reproduce the binary and prove that it matches what the build server builds.
[1588.36 --> 1590.20] So you can prove there is no backdoor.
[1590.86 --> 1597.16] So what I plan to do is to do the first experiment, reproducing the builds of Caddy, but then build
[1597.16 --> 1604.02] some small tools to allow anyone to reproduce builds and publish signatures, maybe with key
[1604.02 --> 1606.84] base to show that they match.
[1606.98 --> 1610.60] Maybe even publish them to a key transparency log, which is like CT.
[1610.98 --> 1613.20] But now I'm crypto nerding too hard.
[1614.16 --> 1615.10] You lost me there.