[09:36] <tomwardill> ricotz: builders are back (broken image service in one DC, broken image in another)
[09:46] <ricotz> tomwardill, great :), thank you
[11:28] <ricotz> hi, could someone please take a look at https://answers.launchpad.net/launchpad/+question/697119 ?
[11:38] <cjwatson> ricotz: done
[11:38] <ricotz> cjwatson, thank you very much
[14:07] <athos> I am wondering what the way forward on https://bugs.launchpad.net/launchpad-buildd/+bug/1702130 should look like. This one is about using resolver-ant-tasks for snap builds (I am having issues with how the build handles/uses the proxy)
[14:09] <athos> This was for a cassandra snap. Cassandra recently (on cassandra >= 4) moved away from maven-ant-tasts in favor of resolver-and-tasks, and I am wondering if this would be enought to build a new cassandra snap :)
[14:11] <cjwatson> So the proxy is just provisioned through the environment
[14:11] <cjwatson> http_proxy / https_proxy
[14:12] <cjwatson> And it is at least partially being used here
[14:12] <cjwatson> [10/May/2021:13:38:53 +0000] "CONNECT repo1.maven.org:443 HTTP/1.1" 200 4335153 "-" "Java/11.0.11"
[14:12] <cjwatson> ^ that line is logging from the proxy
[14:13] <cjwatson> This is probably a bug in either the snapcraft integration for resolver-ant-tasks (or similar), or in resolver-ant-tasks itself
[14:14] <cjwatson> Easiest way to tackle this sort of thing tends to be to set up a simple squid proxy or whatever, and a container that only has internet access through the proxy and not directly, and try running the build in that
[14:14] <cjwatson> Oddly, for a language that has historically had high adoption in corporate environments, lots of Java code seems to be weirdly bad at handling proxies
[14:15] <cjwatson> But LP staff are mostly not Java programmers so we really need help from actual Java people to sort this kind of thing out
[14:16] <ogra> if you just need to transition to a different env var dynamically based on http_proxy being set you can do that easily from inside snapcraft.yaml though ...
[14:16] <cjwatson> What does "transition to a different env var" mean here?
[14:16] <ogra> bug #1886861 has it in the description
[14:17] <cjwatson> Ah right.  This may or may not be the same thing
[14:17] <ogra> oh, and sergio actually talks about maven in there
[14:17] <cjwatson> I spent *days* going down a rabbit-hole of things I didn't understand at all last time I tried to debug maven-ant-tasks
[14:17] <cjwatson> And didn't really get anywhere
[14:18] <cjwatson> So I'm quite strongly inclined to explain to a Java person (e.g. athos, I hope) how the system works in general and let them figure it out :)
[14:18] <ogra> well, it got assigned to chris 🙂
[14:18] <cjwatson> Can we avoid conflating npm and maven though please?
[14:18] <ogra> probably just fell off the table because nobody made any noise on it anymore
[14:19] <cjwatson> The underlying cause is probably completely different
[14:19] <ogra> k
[14:19] <cjwatson> I don't know why Sergio randomly brought up maven in that bug
[14:20] <cjwatson> But Java and JavaScript are not the same and don't have much to do with each other :)
[14:20] <ogra> indeed
[14:20] <cjwatson> It might indeed be that resolver-ant-tasks needs some more environment variables or manually setting some Java properties or ... I don't know
[14:20] <ogra> and this is even specific to npm ...
[14:22] <ogra> (not even javascript, just to enable its scripts to download tarballs)
[14:27] <athos> cjwatson: ack; I will get squid in that container and start debugging the issue then (btw, I am not a java person)
[14:34] <cjwatson> OK, I was just hoping :)
[14:34] <cjwatson> squid would most easily be outside the container FWIW
[14:35] <cjwatson> That way you can easily prevent everything in the container from any internet access other than squid, and squid can talk freely
[14:35] <cjwatson> (easily> with ufw or similar)