IRC channel logs


back to list of logs

<taeaad>What does this error mean? "ERROR: 1. &test-system-not-found"?
<meaty>is there no 'gsettings' package in guix?
<meaty>nvm, a reddit thread says it's in glib:bin
<meaty>another thing on my to-do list ig
<freakingpenguin>Oooh, the tweak to Cuirass's ui for evaluation lists look pretty sharp. It also seems more responsive?
<trisquel-49b5>is there like anything like flakes in guix
<trisquel-49b5>and yes using liveusb of anoher distro
<trisquel-49b5>just testing the sugar ui ignore that
<picnoir>cbaines: do you have any special trick to prune/gc the cached NARs from a nar-herder mirror instance or do you use a LRU-based policy?
<picnoir>IE. do you have a feedback loop between the narinfo DB deletions and the stuff cached in the companion nginx instance?
<cbaines>picnoir, for the mirrors I operate, the caching is managed by NGinx
<cbaines>currently the NGinx cache isn't changed if a nar is removed, but that's something that could be added
<picnoir>right, thanks!
<taeaad>At, there is a script,, do I need to recreate this script in my manifest, or should I call it? I'm creating a python-lightgbm that inherits from lightgbm (similar to python-xgboost and xgboost in machine-learning.scm).
<kks>Surprised that guix's blender is such a low version
<kks>Nevermind, I guess it's the LTS edition?
<Kabouik>wigust: that seems to be an issue with ungoogled-chromium. If I use Icecat, I can see the virtual camera as an input in the webcam tester, and also in the configuration test of the Jitsi-based videoconferencing app I use the most (but not when actually in a conference, just in the testing part, I wonder if this may be due to a resolution limit on webcams). Surprising that it would work better in Icecat, I wonder if this might be due to hardware
<Kabouik>acceleration in Chromium or something. I usually have much fewer things working in Icecat out of the box compared to Chromium.
<taeaad>How can I get the SHA256 identifier for an arbitrary git repo, given a fixed version?
<taeaad>This? git clone; cd foo; guix hash -x --serializer=nar
<taeaad>I think I need to specify which file to hash.
<Altadil>Hello, when trying to send a patch to guix, with git-send-email, it seems the address to send to is no longer automatically filled. Is this happening only for me?
<taeaad>Is it the base32 of the commit hash?
<cbaines>civodul, have you seen ?
<peanuts>"[PATCH] store: Refactor connect-to-daemon."
<taeaad>I'm having trouble creating my own python-lightgbm package. Can I just wrap guix around the Python wheel file? It installs easily with Pip called from Guix Python and works when I import it.
<cbaines>I'm having blocking syscall issues with the build coordinator, and I mostly want to eliminate with-store/non-blocking as a cause
<lechner>Hi, what's the Mumi URL to download the first (and only) patch-set from report #54165, please?
<peanuts>"[PATCH] gnu: guile-bytestructures: Update to 2.0.1."
<lechner>nvm, i missed an '/issue' path component
<cow_2001>waiting for zipheir to merge it into his branch. i've added docstrings from the SRFI document to the procedures and added a testing script for guile.
<peanuts>"kakafarm/guile-srfi-189 -"
<peanuts>"SRFI 189: Maybe and Either: optional container types"
<reyman>hi guixer !
<reyman>i'm trying to mount.nfs as user, the mountpoint is correctly set in my /etc/fstab, setuid is done on mount.nfs ... but when i run "umount.nfs ~/srv/blackhole/" i have a permission denied: no match ...
<reyman>if you have a nousermy fstab is /home/reyman/srv/blackhole/ nfs4 rw,noauto,user
<reyman>my fstab, sorry is " /home/reyman/srv/blackhole/ nfs4 rw,noauto,user"
<reyman>i'm interested if you have a mount.nfs that run without sudo
<freakingpenguin>I noticed that with mount.cifs earlier. Even though it was setuid it needed to be run as root otherwise it would inexplicably complain about fstab. Not sure why.
<reyman>ok so i'm not alone :D
<reyman>i will try to overwrite folder uid/gid during mount but this is not the simple way to do
<Guest81>I just installed Guix and my cursor disappears when I log into a session. The cursor is present in GDM, but it's completely invisible in EXWM. i3 shows the cursor for about 10 seconds before it disappears for good. What gives?
<Altadil>Guest81: I know I’ve had an invisible cursor with X sessions and had to switch to wayland to solve it. Maybe it’s the same issue?
<fnat>I see 'guix deploy' has no '-e/--expression' option. Could it be a good idea to add that?
<fnat>Other than making it consistent with other Guix subcommands, it'd make it easier to have the deployment info in the same file as the operating system definition. That way, one can simply call 'guix deploy --expression=(...)' instead of having a separate deployment file.
<Guest81>Altadil: Well that's a shame! I wanted to experiment with EXWM but it's going to be a lot harder if I can't see my cursor. Technically you don't need a mouse but a lot of graphical apps I use don't really work in a keyboard centric interface
<Altadil>Guest81: be sure to check this hypothesis first, though. I don’t know for sure. :)
<Altadil>But indeed, if that’s the same bug and exwm doesn’t have a wayland session, that’s a shame. :(
<fnat>Guest81: EXWM user here, the cursor looks fine, maybe there's something you can tweak in your configuration. I know I'm not very helpful here, but just to let you know that it works under the proper settings.
<fnat>Guest81: Let me see if there's anything specific in my config.
<fnat>Guest81: Hm, weird, no, it doesn't seem so. The only thing that comes to mind (although it's unlikely to have any role in this?) is: '(modify-services %desktop-services (delete gdm-service-type))'.
<Guest81>The peculiar thing is that the cursor stays momentarily visible in i3. It doesn't update, but it's visible for a few seconds
<fnat>Guest81: That'd make me think of a decluttering service that hides the cursor when not used, but it's gone rogue somehow? :)
<fnat>A very weak lead, but that's why I was thinking of '(delete gdm-service-type)'.
<fnat>Other than that, I can't see anything specific in my system config. And the cursor is visible, at a eye-comfortable size.
<zeropoint>Guest81: do you have hardware acceleration activated for the cursor? I had to disable that (using x) in order for my cursor to show up.
<Guest81>zeropoint: I wouldn't know, I've never used X. All of my experience has been on Wayland.
<Guest81>How exactly would I check?
<zeropoint>I don't know wayland, but X has an option to disable hardware acceleration
<Guest81>I am using X, how do I disable the hardware cursor?
<zeropoint>here's the relevant portion from my system config:
<zeropoint>see SWCursor option
<peanuts>"debian Pastezone"
<ieure>I haven't needed to disable X11 hardware pointer stuff since like the 1990s. It used to be very fickle.
<Guest81>The plot thickens, I decided to try a wayland session as Altadil suggested and the cursor is visible. I'm going to give SWCursor a try since I really want to try EXWM
<Deltafire>does suspend/standby work for anyone else?
<ieure>It works for me.
<Deltafire>i mean, i guess it does work - it's the resuming bit that's not working properly
<Deltafire>interesting, i wonder why it works for you and not me
<ieure>Different hardware? Works on various ThinkPads I have around, X13 Gen 2 AMD, L390 Yoga, P53s.
<freakingpenguin>Is there a reason Guix modules prefer "#:use-module (gnu packages foo) ..." instead of "(use-package-modules foo ...)"? Why the preference for the verbose option?
<freakingpenguin>Functionally does encapsulating imports in the (define-module) declaration give you anything?
<ieure>freakingpenguin, Can't speak for anyone else, but I have my own channel and maintain some Guix packages, I'll usually update my channel before sending patches upstream. #:use-module lets me easily switch whether a package is coming from upstream or my channel; use-package-module only works with stuff in (gnu packages ...).