IRC channel logs

2026-06-20.log

back to list of logs

<randomguixuser>guys! i'm trying setup a container with postgres, like this https://guix.gnu.org/cookbook/en/html_node/A-Database-Container.html, but i have a weird error that says `exception caught while starting postgres: (%exception #<&invoke-error program: "/gnu/store/xsfj3hq0zzhvs86pxqcvvp9rwa39j4sx-pg_ctl-wrapper" arguments: ("start") exit-status: 1 term-signal: #f stop-signal: #f>)`, i don't know whats happening. initially i was trying with my own c
<randomguixuser>onfig, but even just coping and pasting the docs example as is i have this error... any ideas?
<ieure>randomguixuser, Anything in the logs?
<randomguixuser>ieure: i have this
<randomguixuser>2026-06-20 00:22:43 localhost shepherd[1]: Starting service postgres...
<randomguixuser>2026-06-20 00:22:43 localhost shepherd[1]: [3gzm0lrxqnhpgqgw65jjxjaiqx6q7r45-pg_ctl-wrapper] Backtrace:
<randomguixuser>2026-06-20 00:22:43 localhost shepherd[1]: [3gzm0lrxqnhpgqgw65jjxjaiqx6q7r45-pg_ctl-wrapper] Exception thrown while printing backtrace:
<randomguixuser>2026-06-20 00:22:43 localhost shepherd[1]: [3gzm0lrxqnhpgqgw65jjxjaiqx6q7r45-pg_ctl-wrapper] In procedure public-lookup: Module named (system repl debug) does not exist
<randomguixuser>2026-06-20 00:22:43 localhost shepherd[1]: [3gzm0lrxqnhpgqgw65jjxjaiqx6q7r45-pg_ctl-wrapper]
<randomguixuser>2026-06-20 00:22:43 localhost shepherd[1]: Service postgres failed to start.
<ieure>Seems like a bug.
<randomguixuser>that is in `/var/log/messages` but in `/var/log/postgres/*` there is no any ouput.. can be `no module system repl debug` a bad config of me?
<randomguixuser>a missing dep/config or something like?
<sienwick>Test
<mlxdy>Compilations are often in Guix? I wanted to install Tor Browser and I have to compile it
<sienwick>anyone know a good terminal based matrix client?
<sienwick>thats supported on guix
<csantosb>Hi Guix ! Funny thing: `guix shell -CN --preserve='^DISPLAY$' sby-gui -- sby-gui` opens a gui, as expected
<csantosb>Dropping the `N` network produces a `qt.qpa.xcb: could not connect to display :0`
<sham1>That is rather odd. You'd think that the networking capability wouldn't affect unix sockets
<csantosb>Something similar happens with `guix shell -C ungoogled-chromium --preserve='^XAUTHORITY$' --expose="${XAUTHORITY}" --preserve='^DISPLAY$' -- chromium`
<futurile>csantosb: does it happen with other QT apps? I'm guessing it's localhost
<csantosb>With `guix shell -C --preserve='^DISPLAY$' appcsxcad -- AppCSXCAD` I see the same problem
<futurile>wow interesting
<csantosb>I noticed that as, when building a docker image with sby-gui, I need to run it as:
<csantosb>`podman run -it --rm --network=host -e DISPLAY container sby-gui`
<futurile>so I gues the underlying linux capabilities for 'network' cover something that X Windows requires!?
<futurile>and it's not specific to our implementation in guix shell if podman does that
<futurile>I often think there's so much in Scheme but it's hidden behind a wall of academicy language with no actual examples of anything anyone would actually want to do
<futurile>mostly because it's written by academics who can't help making themselves sound clever, even when the _thing_ is quite clever enough
<futurile>(goes back to reading an srfi)
<cow_2001>how do i clone git submodules in a git-origin?
<look>cow_2001: (git-reference ... (recursive? #t))
<mlxdy>I have mate service, how can I use Mate without gdm?
<cow_2001>oooh
<cow_2001>look: thank you!
<mlxdy>I tried to add lightdm-service-type, but I can't because xorg-server provided more than once
<mlxdy>I removed gdm service already
<cow_2001>oh no. i've only skimmed until now. actually reading i see it says "will fetch automatically on demand" https://github.com/riscv-collab/riscv-gnu-toolchain says its compilation
<cow_2001>err
<cow_2001>anyway, it needs a local git repository for it to fetch the submodules automatically, but guix doesn't keep the .git, just the worktree
<cow_2001>in the .git there are all the references to the submodules
<look>mlxdy: can you post your config somewhere like debian paste? it would be easier to help you debug
<mlxdy>Yeah, moment
<cow_2001>and the (recursive? #t) bit did not work
<mlxdy>look: https://pastebin.com/dgCBWRe3
<mlxdy>I added these mate packages because my screenlocking dont work and I can't edit my menu
<cow_2001> https://paste.debian.net/hidden/c6839d3a
<cow_2001>sorry, didn't include the use-modules
<cow_2001>here you go https://paste.debian.net/hidden/3b27b6aa
<cow_2001>(it's not expat licensed)
<mlxdy>After I made system reconfigure wifi stopped working at all
<mlxdy>gdm is still loading
<mlxdy>cat stopped working
<sham1>You shouldn't need to list the mate packages when using mate-service-type. That's very peculiar
<mlxdy>Okay, I'll change that.
<sham1>Also, to call `set-xorg-configuration` when you want to use lightdm-service-type, you need to tell it that. So it's something like `(set-xorg-configuration xorg-config #:login-manager-service-type lightdm-service-tye)`
<sham1>See: https://codeberg.org/guix/guix/src/branch/master/gnu/services/xorg.scm#L1421-L1434
<cow_2001>look: can you spot why it won't fetch the submodules?
<mlxdy>sham1: I don't get how to do this what you said
<look>cow_2001: seems to be fetching for me, a sec, building the checkout rn
<cow_2001>the problem, when i (invoke "find") in the added phase, i see that the directories of the git submodules are empty
<cow_2001>oh, another thing i want to do is fetching the git submodules shallowly
<look>still building the checkout lol
<cow_2001>?!
<cow_2001>look: can't you shallow it?
<cow_2001>looking in the manual i don't see it
<look>I don't think so, I tried reading the source but it does some submodule magic with (git submodule)
<look>seems like the submodule logic does not respect .gitmodules preferences
<look>(although the file itself is never deleted)
<cow_2001>i might have to first do it manually and then point at a local file source?
<cow_2001>oh my god. what were they thinking
<look>took me 35mins to checkout, but all the submodules files are there
<cow_2001>and it's 8G :|
<cow_2001>i need a new computer :(
<look>mlxdy: I took a look at your config, the problem you're having is because you can't pass the xorg service directly with lightdm-service-type, because it already extends it
<look> https://paste.debian.net/hidden/3d78ea63 try this
<look>lightdm has a field to pass a xorg-configuration directly
<mlxdy>Thank you! I had much more problems. Wifi stopped working and Mate wasn't fully functional (even after I used the same config as before my modifications with login manager etc.). As I need to have tommorow working system I need to go with something else at the moment. I can't trust Guix that much.
<cow_2001>good lord
<ieure>mlxdy, Did you have a working config earlier? You can roll back to that system generation, or reconfigure using the old configuration, until you get things sorted. `guix system list-generations' shows what you have, `sudo guix system switch-generation' will change.
<ieure>mlxdy, Each entry in the output of `list-generations' points to the config it was built from, you can `sudo guix system reconfigure' using any of them to get a newer version of that config.
<mlxdy>Thanks, I know that I can rollback my system. But Guix don't look as an option right now.
<z572>civodul: Can we rebase the core-packages-team? I plan to upgrade the default gcc to gcc16 before the next release. see https://codeberg.org/guix/guix/pulls/9386
<civodul>z572: sure!
<civodul>do you want to do it yourself?
<civodul>i haven’t been able to do meaningful work on this branch over the past couple of months :-/
<civodul>we have a bunch of nice changes that are queued and waiting to be merged
<civodul>(by “do it” i was referring to rebasing)
<z572>civodul: ok, i will rebase it, and backup original core-packages-team to core-packages-team-2026-06-20
<civodul>z572: great, thanks!
<Rutherther>is there any plan on when core packages could be ready?
<civodul>no, unfortunately
<civodul>yelninei and others have been doing great work but now we should work on stabilizing it
<Rutherther>I think it would be nice to time the next release so that it happens after core-packages merge
<Rutherther>anyway if it was taking too long, probably better to not put off the release due to it
<civodul>yeah
<civodul>but i guess we can just agree to stop making changes to it anytime
<civodul>it’s “just” that someone need to be taking care of it
<civodul>and i’m the only committer maybe in the core-packages team?
<civodul>ah no, actually we’re all committers
<civodul>so probably mostly a coordination problem :-)
<z572>It seems that some of the 88d75f99c35ae1dae2bec2ebe9229d7c7c988dea have been merged into master. See https://codeberg.org/guix/guix/commit/4b5d81fa9d4673f4d40799243eebe051a7f95e54. I'm not familiar with Haskell, so I'll skip this for now.
<z572>and skip 98683348d70f0e105955bdf293b5e31abbd31408 too for now.
<z572>civodul: see https://codeberg.org/guix/guix/src/branch/core-packages-team-2026-06-20-rebased , If there is no problem with this, it will be force-pushed to the core-packages-team branch.
<z572>GHC modification was restored and placed at the end.
<z572>Merge the commit have "follow-up" into the previous commit, and adjust message.
<z572>civodul: force-pushed, `./pre-inst-env guix build hello` and `./pre-inst-env guix build --target=x86_64-pc-gnu hello` build success for me
<jlicht>csantosb: emacs-orglink does not build for me after the bump, did it somehow work for you?
<csantosb>jlicht: let me check
<csantosb>I see. It requires llama and newest org. Let me fix it.
<jlicht>csantosb: thanks, always good to know it's not me being crazy :)
<pdpnw>moved over to guix after using arch for over a decade, really enjoying it so far. wish I'd had something like this back in my infra days
<sham1>I mean, Nix was already a thing back then, but yeah. Hell, I think that even guix was already a thing back then
<SillySpider>After running `guix pull` as a foreign package manager, pull is hanging while the last string output is `substitute:`. That's after downloading a few substitutes successfully. I'd like to try `guix pull --no-substitutes`, so is it safe to SIGINT pull to end the hang? The last time this happened, I embarrassingly FUBAR'd my guix by manually deleting some files that I *thought* were just cache files
<SillySpider>(mea culpa), so I'm trying not to assume I know more than I do, this time.