IRC channel logs
2026-01-17.log
back to list of logs
<meaty>is there a way to quickly try out a system service without reconfiguring the system? <meaty>I know there's probbably a way to hack it with virtual machines, but I also know there's a REPL somewhere in shepherd that might have a way to do something? <meaty>I found it shepherd's repl, type ",bournish" <psycotica0>I don't know, but I might _guess_ that it's something that acts kinda like "bourne shell" / sh, and is thus "bourne-ish"? <yarl>Who can set labels? on issues on codeberg? <sneek>Welcome back linus_, you have 1 message! <sneek>linus_, rlb says: hah, well whatever that was on facebook it wasn't me -- I've never been there, and glad scheme/guile's worked out for you. <civodul>old: guile 3.0.11 is on ‘next-master’ <civodul>will be merged after the release, hopefully in just a few days <zenmaya>did anyone encounter the issue that networkmanager is getting suspended by elogind (setting interfaces to unmanaged) and never waking them up afterwards? <zenmaya>this is happening to me and it means that i have no network connection after a suspend <zenmaya>this did not happen on gnome, but i moved to niri, and now i have to restart NetworkManager service every time <old>civodul: Do you think it's possible to keep 3.0.9 around at the same time? I know that 3.0.11 fixed srfi-64, but it actually breaks tests that were written in 3.0.9 <old>I figure that out thanks to arch-linux and made myself a thin wrapper that keeps things compatible between broken and fixed srfi-64, but I figured that some Guile project might not and will have a big surprise once 3.0.11 is merged <Deltafire>zenmaya: I'm not encountering that issue, also using niri and nm <zenmaya>Deltafire: interesting, do you mind sharing your config? <civodul>old: yes, we’ll definitely keep 3.0.9 around for a while <civodul>i had to fix Guix tests that relied on the (incorrect) srfi-64 behavior from 3.0.9 <civodul>old: sophisticated! for Guix i just adapted to 3.0.11, and that also works with 3.0.9 (was ‘test-error’ mostly) <old>ahh yes. I've mostly avoi test-error in blue because it was broken <civodul>(BTW, i’m glad i’ll soon know more about BLUE :-)) <old>ACTION have to make the slides .. <civodul>Rutherther: i’m testing the manual redirects, which are now deployed on berlin <civodul>ACTION -> dinner, enjoy the release meeting! <attila_lendvai>i managed to set up hibernate, and it works by "echo disk >/sys/power/state", but gnome doesn't show the icon. presumably because it's not compatible with elogind? should i be content with an echo, or can gnome be tamed on guix? <yelninei>what do you mean by icon? afaik there is only a extension for it. Also loginctl "hibernate hibernate/hybrid-sleep" work for me <yelninei>* loginctl hibernate / loginctl hybrid-sleep <attila_lendvai>yelninei, sorry, i mean button. next to where the shutdown/restart are. <yelninei>why this is not available by default I dont know <Wurt>Hello! I ran `guix lint' on a package that has the input `(list glib "bin")', and the linter suggested that I fix the input name. Is this a bug, or do I need to reference it another way? <ieure>Wurt, It should be `(,glib "bin") <ieure>`(glib "bin") -> (list 'glib "bin") ;; not what you want <ieure>Issue here is that I need python in the profile I'm building in that field serializer, but the correct Python to use is at the top of the configuration hierarchy. <ieure>Anyone have a suggestion for approaches to make that context visible in the scope of the field serializer? <ieure>The two idea I had were to add a `python' field to collectd-plugin-python and rewrite the configuration prior to serializing, copying it from the top level; or to make a parameter, then a `serialize-collectd-configuration' procedure which binds it, so the lower-level field serializer can get at it. <ieure>But I don't like either, they're pretty hacky. <ieure>I guess the first one is slightly less bad. <Wurt>ieure, I use backticks to mark code, I shouldn't use it. My input element is (list glib "bin"), that I think is equal to `(,glib "bin"). <nckx>Did the linter sot suggest *how* to fix it or what it expects? <nckx>(That would be a 'bug' in itself.) <Wurt>nckx, guix lint output is: /home/wurt/src/guix/gnu/packages/video.scm:7071:19: showmethekey@1.19.0: label 'glib' does not match package name 'glib:bin' <ieure>Wurt, Yes, (list glib "bin") is equivalent. <ieure>Wurt, I think either the linter warning is just buggy, or getting confused and thinking that your alternate-output input specification is the old-style labeled one. <ieure>Wurt, Used to be that you'd do ("some-alias" glib), and you could then (assoc-ref #$inputs "some-alias") to get stuff out of it. <Wurt>ieure, so should I ignore it? Thank you! <ieure>Wurt, Ignoring is probably the right move, but you could paste your package definition if you want a human linter pass. <ekaitz>csantosb: Could not determine wxWidgets version. Python plugins will not be available. -- This thing in kicad, did you see this before <csantosb>ekaitz: nope, but you already had problems with that when first packaged kicad, see wxwidgets-sans-egl <csantosb>python-wxpython (a dependency of kicad), for example, uses wxwidgets <Wurt>Why does guix style prefer a one-space indent instead of two right after git-reference? Emacs does not auto-indent like that. <civodul>Wurt: that looks like a discrepancy between .dir-locals.el and %special-forms in (guix read-print) <Wurt>civodul, which indent should I choose? It doesn't care? <ekaitz>civodul: about that I'd say we should have a flatter indentation system, i don't think keeping those rules is accessible neither desirable <ekaitz>csantosb: i'll take a look... i'm pretty annoyed by that message <ekaitz>maybe I just need to upgrade my system, i tend to use old versions all the time <csantosb>wxwidgets-sans-egl comes from the 9.0.0 ages, it's been a while and maybe this is not necessary anymore <yelninei>attila_lendvai: You could try packaing as a guix package or install it locally: look at what the makefile in the project does <yelninei>ACTION has successfully rebuild the hurd manifest twice on next-master <yelninei>(apart from the things that are currently broken, but at least no regressions) <yelninei>civodul: did you see my comment on the loss of the sigprocmask for shepherd/shepherd#80 which makes shepherd kill itself instead of the child