IRC channel logs

2025-03-18.log

back to list of logs

<cwebber>o/
<cwebber>hi hi
<jfred>ACTION waves
<dthompson>hey cwebber and jfred
<dthompson>jfred: tsyesika is really on top of things and made a PR to fix nested with-vat https://codeberg.org/spritely/goblins/pulls/668
<jfred>oh whoa that was fast!
<dthompson>yeah tsyesika is working at light speed on a goblins patch release with lots of little fixes
<jfred>I spy with my little eye... a new pumpkin! https://codeberg.org/spritely/pumpkin-chat
<dthompson>jfred: haha yes this is our latest dogfooding experiment
<civodul>oh Spritely on Codeberg, sweet :-)
<dthompson>civodul: working on it! been "almost done" for days now
<civodul>neat!
<dthompson>on the final step of getting our forgejo-runner instance deployed
<civodul>hence forgejo-runner i guess :-)
<civodul>cool
<dthompson>yup :)
<dthompson>I need to take a breather once this is done but I want to get all the packages for forgejo-runner upstream
<civodul>that would be great
<civodul>Cuirass is gaining support for Forgejo too, in case that can be of any help
<dthompson>the service I'm deploying is very barebones, just enough for our needs, but I'd like to get that upstream, too
<dthompson>civodul: oh? like to be used as a runner?
<civodul>it already implements webhooks, meaning it can create/delete jobsets for pull requests
<civodul>and now Romain is working on notifications: getting feedback in PRs
<dthompson>okay so you skip the whole actions thing entirely
<dthompson>I see
<civodul>it’s actually merged but still WIP :-)
<civodul>right
<dthompson>maybe we should just use that
<civodul>if it’s just for builds and distributing substitutes, it should be a good fit
<dthompson>idk... I should just get this thing running so we can complete the switch
<civodul>if you need other things, actions are better
<civodul>yup
<civodul>anyway, happy to discuss it again as both of us make progress
<dthompson>yeah mainly we want to know that 'make check' and 'make distcheck' pass, but sometimes we want other things.
<tsyesika>we also check my emacs isn't polluting our codebase with tabs ;)
<tsyesika>(which it has many times for some reason)
<dthompson>hoot and goblins both have some additional jobs that integrated with gitlab features. I haven't yet seen if we can port those over yet
<civodul>heh :-)
<civodul>yeah for things like linting, forgejo-runner is a better fit
<dthompson>tsyesika got our test suite to generate junit xml which gitlab understood to give detailed test suite feedback
<dthompson>the hack I'm doing with forgejo-runner is skipping the docker stuff
<tsyesika>yeah we also had some performance stats, I haven't actually checked if forgejo supports anything like that
<dthompson>I wish it were easy to extend the runner with new runtime modes, so I could have a properly integrated 'guix shell' mode
<dthompson>but, as a hack, running without isolation and using 'guix shell --container' after cloning the repo seems to work well enough
<civodul>ACTION nods
<dthompson>what I'm saying is that someone (not me) needs to write a forgejo-compatible runner in guile that has guix integration :)
<civodul>heh, that too would be nice
<dthompson>I will assign this task to jfred
<dthompson>good luck, jfred. I believe in you.
<jfred>😆
<dthompson>civodul: while I have you here ;) what should I look at in shepherd if a new service I expect to see in 'herd status' doesn't show up after a system reconfigure?
<dthompson>I don't see an error in /var/log/messages about it, but I assume my service code is broken in some way
<dthompson>well, I found the error. shepherd didn't help me, though.
<Gooberpatrol66> https://youtu.be/D0wbCctcEIA
<Gooberpatrol66>a youtuber made a video from dustycloud's bluesky article
<dthompson>Gooberpatrol66: whaaaat thanks for the link