IRC channel logs

2024-11-27.log

back to list of logs

<PotentialUser-78>Hi, I'm trying to package an app, but the build system needs to make HTTP requests, and it seems that guix has a network sandbox. Is there anyway I can get by this so the HTTP requests can go through?
<gabber>PotentialUser-78: short answer: no.
<gabber>the idea is to "clean up" the building process: get all dependencies through usual guix fashion and then build the software without the HTTP requests
<PotentialUser-78>I see. I assume that's why for the cargo-build-system you have to package every dependency and list them
<PotentialUser-78>I'll see if I can do something like that for this. Thank you for the response
<gabber>exactly - HTH
<futurile>PotentialUser-78: are you trying to package an app FOR Guix, or for your own use?
<wolfdog>is there any reason for the emacs' geiser-chicken package (from MELPA) not being in Guix? or is it just a "not packaged since no one has needed it in guix" kind of case
<futurile>PotentialUser-78: if it's for your own use, you can do whatever you want
<gabber>wolfdog: more probably: not packaged because no one has packaged it yet ;)
<mange>You could take a look at https://github.com/garrgravarr/guix-emacs if you want up-to-date melpa packages, though.
<PotentialUser-78>futurile: my own use, but I'd like to do it the Guix way, as I know some people who also would like to use the app with guix
<noe>Out of curiosity, what’s the build system?
<wolfdog>hm, seems like it fails to build with the current geiser version (with a package definition generated with `guix import elpa -a melpa geiser-chicken'): https://paste.debian.net/1337035/
<wolfdog>oh welp
<PotentialUser-78>noe: zig-build-system
<futurile>PotentialUser-78: if it's your own use, you could (a) use a guix.scm (rather than create a package), (b) create a package (but build it in a guix shell or use time-machine so it's pinned)
<futurile>PotentialUser-78: as you're building the package yourself, you can have 'guix shell/guix time-machine' have access to the http if that's what you want
<futurile>PotentialUser-78: the reason we can't do that in the distro is we want everything to be fully reproducible - same on Debian - but for your own packages it's totally fine
<futurile>PotentialUser-78: I think Dave Thompson's blog has a post about it: https://dthompson.us/posts/guix-for-development.html and I wrote a couple about it as well on futurile
<PotentialUser-78>futurile: Alright, thank you for the information
<divya>What's the way to change display managers from gdm to sddm? I removed gdm from desktop-services and added sddm-service-type but on login I don't sew anything.
<gabber>divya: WDYM? the screen stayed black?
<gabber>sddm / gdm are the "login screens" where you enter user/password
<divya>gabber: Yes, and it shows login prompt on TTY.
<divya>I went to TTY7 with C-M-7 but it only shows a blinking underscore
<gabber>interesting! is there anything of interest in the logs (/var/log/messages) ?
<divya>Remind me again where should I be looking for the logs? in /var/log/ there's only gdm and guix
<gabber>there's no /var/log/messages ?
<gabber>ACTION is going afk
<divya>gabber: https://0x0.st/XRHW.txt
<divya>And here's the config I'm trying: https://0x0.st/XRH4.scm
<jmes>Anybody know the hardware specs for any of the guix build servers? I have multiple guix users at home and we're thinking of building our own build server for the LAN.
<jmes>I know it doesn't matter much. My desktop is powerful enough itself but I'm cursious about the specs of the substitute servers.
<jmes>s/cursious/curious
<lfam>jmes: The bulk of the compute is AMD Epyc servers with 96 cores and a large amount of memory
<sepeth>Hey Guix, I am trying to upgrade rust-webbrowser, and I changed the current version to the next version and inherit the current version def from the next version via (package (inherit ...)), but I am getting error: name: unbound variable and such errors.
<sepeth>And it ends with this error:
<sepeth>Throw to key `match-error' with args `("match" "no matching pattern" (unbound-variable "resolve-interface" "no binding `~A' in module ~A" (shared-mime-info (gnu packages freedesktop)) #f))'.
<efraim>try not removing the name field from the inherited package
<sepeth>"Invalid format for name (no special chars, max 10 chars)" on debian paste is so annoying</rant>
<Rutherther>sepeth could you send the whole expression via paste site?
<sepeth>On it
<sepeth> https://paste.debian.net/1337058/
<sepeth>FWIW, this is on rust-team branch
<efraim>you need to keep the name field for rust-webbrowser-0.8 so it can be used in file-name
<sepeth>Oh!
<sepeth>Good catch, thank you!
<koorosh>hi I need the python idlelib module for a project I installed python:idle but the when I run python3 -m idlelib I get the message  "/gnu/store/yzqiy26lwn6gwlfa5pda6qmx4awdx9lk-profile/bin/python3: No module named idlelib"
<koorosh>is there anyway to fix this I've also tried adding the site package path of python:idle in store to GUIX_PYTHONPATH and PYTHON_PATH
<Rutherther>koorosh: if you are installing python/python module for first time, did you source your profile? / try the same thing in a shell first
<koorosh>yes just tried it with source ~/.guix-profile/etc/profile and then python3 -m idlelib it is still the same thing , running the idle3 binaries installed with python:idle results in the same issue
<sepeth>Workflow question w.r.t. creating a patch file for a package. How do you do it? Is there a guix command that pulls git repo of the package and you modify the package and take the git diff or format-patch output? Or, something else?
<Rutherther>koorosh: and you have also python(:out) installed in the same profile?
<koorosh>Rutherther: yes
<Rutherther>Okay, then I dont know whats wrong as I am currently not near a machine with guix to debug it myself
<koorosh>Rutherther: tnx for the help anyway I come back here and ask around
<sepeth>koorosh, with python-next I see idle in the store, but not with current python
<sepeth>So I did this,
<sepeth>guix shell python-next
<sepeth>PYTHONPATH=/gnu/store/*-python-next-*-idle/lib/python3.12 python3
<sepeth>Then import idlelib worked.
<sepeth>I am not sure of a way without the PYTHONPATH trickery. "idle" seems to be a separate output here: https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/python.scm#n689
<koorosh>sepeth: tnx alot i'll try that
<koorosh>that worked for me also doing the same with the normal python package worked as well
<koorosh>what is the difference between GUIX_PYTHONPATH and PYTHONPATH how does python use the GUIX_PYTHONPATH
<sepeth>Good question, but idk :(
<sepeth>Maybe there's a discussion somewhere in guix devel, but couldn't find it from the git blame NEWS or other places. I wish guix commits did add some context around whys. Commit messages usually just say what, not why.
<Rutherther>koorosh guix_pythonpath is put to pythonpath eventually. Its for separation of variables between guix packages and other stuff like venv, so that you dont override pythonpath when you dont want to.
<koorosh>Rutherther: thank you
<divya>A day of trying to boot Guix, now I see in /var/log/messages that amdgpu isn't starting for some reason :)
<koorosh>divya: i think you might need some firmwares take a look at nonguix
<divya>koorosh: Exactly what I was about to do. But why? I was told to buy AMD because they have everything as free software
<unmush>whoever said that probably didn't consider firmware to be software
<divya>ugh
<divya>I thought I could live with linux-libre :/
<divya>the thing is it works sometimes, when I do a fresh install it somehow works and boots as usual. But when I do guix pull update and make a few changes it breaks
<divya>Could a BIOS update fix anything?
<catsquotl>Anyone know how to remove all the info entries in 6 different languages and set the english one as default? I seem to always open the spanish version by default and all the other language entries are clogging up my info top dir. Removing them from /usr/shar/info/ does not seem to work. I am guessing guix stores them somehwere else and builds the
<catsquotl>top dir from that?
<efraim>it might be possible to modify the profile hook which generates the info-dir-file to only choose certain languages, but I've never looked at doing something like that
<futurile>divya: super annoying, but it *does* sound like you're making some progress!
<apteryx>what does "Be careful about using '$' to match the end of a line; by itself it won't match the terminating newline of a line." really mean in substitute* docstring?
<apteryx>I see it used in an example, but when I try to use it myself, it indeed will cause the pattern to not match, although the same pattern in regexp-builder in emacs works
<apteryx>(the example is in the same docstring)
<apteryx>seems safer to always use .* instead of $ in the substitute* pattern, even when the later would be more correct
<apteryx>or maybe /some-pattern/$.* to match the whole line including the newline
<futurile>apteryx: I spent time this week playing with substitute* and Guile regexp, you can use $\n or $[[:whitespace:]] to get the newline. Also, as substitute* is only matching a line at a time (it doesn't do multilie matching) I found it easier to match on the things on the line and just leave the line break there
<futurile>apteryx: I guess you're trying to completely remove the line, including the line break
<erik``>Hi! Is it possible to specify the channels for a manifest, so that one does not need to invoke time-machine for it to use a specific revision?
<futurile>erik``: AFAIK, no - you have to use time-machine with a manifest and a channels file
<futurile>erik``: I guess you could create a script to invoke them together if you wanted to simplify it
<erik``>In that case, I will just stick to having them invoked together. I guess I could use an inferior, but I have understood that inferiors are still a bit experimental. Thank you!
<futurile>erik``: you might want to ask on the mailing list - explain the use-case/situation - I wonder if it's something that someone would like to add as an option to guix shell etc - no idea what Nix does
<bost>Hi. Can anybody explain me why `guix package -p /home/bost/.guix-profile -I with-editor` returns twice the same result: 'emacs-with-editor 3.4.2 out /gnu/store/gaqc3ad1qw7bcdzhk1hyh92ixiq3bk6k-emacs-with-editor-3.4.2' ?
<janneke>what is a good gnome-terminal alternative? i appreciate its fonts and easiness of font scaling, color schemes, but i'm starting to develop a distaste for its dependency on all of gnome-desktop
<Deltafire>plain xterm?
<Deltafire>anyone know how i can pass arguments to kernel modules, like i would do using /etc/modprobe.d in other distros?
<janneke>Deltafire: great it has decent font scaling nowadays, how does that work?
<Deltafire>janneke: i don't know, i use gnome terminal myself :)
<Deltafire>or 'Console' as it's called in Gnome Desktop
<janneke>Deltafire: yeah, thanks :-/
<erik``>futurile: Good idea, will investigate the approach of Nix.
<efraim>After I switched from enlightenment I switched from terminology to alacritty
<rekado>ACTION takes notes
<efraim>terminology has a couple of cool features, including typop, which will use the terminal to show an image (or video?) as a popup
<Rutherther>erik``: nowadays a lot of people using Nix use flakes, those are containers for Nix codes. In just one file, you specify inputs, these are Nix sources like nixpkgs, and outputs, these are packages, shells etc, made from the inputs.
<Rutherther>erik``: when talking about lower level, you can do in Nix "import builtins.fetchUrl "nixpkgs download" {} ;" directly in Nix file. But its not possible with guix, as the structure is different. Nixpkgs is just Nix sources loaded when evaluating Nix code, whereas guix guile definitions are contained within the guix command itself.
<drocheslav>Anyone succeed in making emacs-guix completion in eshell buffers working? I just have an error "guix-geiser-eval: Error in evaluating guile expression: ice-9/boot-9.scm:1685:16: In procedure raise-exception:
<drocheslav>/home/drocheslav/.config/guix/current/share/guile/site/3.0/guix/scripts/deploy.scm:176:7: Unknown # object: "#~""
<andreas-e>efraim, how is the rust team branch going? I have considered jumping queue with gsl-upgrades since we have not heard back from meson-parallel-tests, but maybe rust should go first ;-)
<efraim> https://qa.guix.gnu.org/branch/rust-team I rebased on a commit QA had processed and now I have numbers!
<efraim> https://qa.guix.gnu.org/branch/meson-enable-parallel-tests no idea for the meson-parallel-tests branch
<andreas-e>In meson-parallel-test there are a few build failures (for instance the tests in gnuradio, so it may be related to parallel tests).
<andreas-e>So it looks like rust is not ready? The x86_64 numbers are quite low on ci.
<efraim>I have no idea why berlin is so low. And I think bordeaux just finished processing the revision yesterday so it can start to build
<efraim>ah yeah, I remember the branch changes table below now, it hasn't started building it yet on bordeaux
<andreas-e>Because meson-updates is first in queue, so it gets blocked in QA, I think.
<andreas-e>The dashboard at ci looks good: https://ci.guix.gnu.org/eval/1850951/dashboard
<efraim>it looks pretty good on x86_64, but it'd be nice to know what QA has to say since it makes it easier to compare branches
<andreas-e>I think it is just not operational right now, we will have to wait for the data service to move to berlin I think.
<maximed-web>gabber: (is the #:target #f flag the indication that cross compilation wouldn't work for a package): no. It tells Guix to not cross-compile it, i.e., compile it as if no cross-compilation would be involved (target=false). Slightly beneficial for data-only packages, and occasionally useful (see e.g. tzdata, where cross-compilation would also lead to
<maximed-web>a problem I don't remember)
<efraim>it's also good for firmware
<andreas-e>efraim, I have just pusged the meson-parallel-tests branch. Hopefully the rust-team branch will be built by QA now.
<catsquotl`>@efraim thanks
<cbaines>andreas-e, efraim I'm guessing that the meson-parallel-tests changes completely overlap with rust-team, so rust-team could probably do with rebasing
<divya>futurile: Yep, it was the kernel. linux-libre doesn't like my new processor
<divya>Also janneke: I'd go with alacritty or if you love patching C apps, suckless' st :)
<apteryx`>andreas-e: thanks for seeing it through!
<janneke>divya: thanks!
<efraim>I'll go ahead and rebase
<efraim>although part of me says I should wait for one of these to turn green first https://data.qa.guix.gnu.org/repository/2/branch/master
<cbaines>efraim, maybe, at least if you wait until the data service starts to process a master revision, then hopefully you'll rebase on one that will be processed
<divya>efraim: Did you patch the blue-recorder patches? I've not been able to get back to my system because of these boot errors, so haven't checked the mails.
<apteryx`>cbaines: I think touching meson/meson-build-system somehow rebuilds the whole rust toolchain
<apteryx`>at least I remember rebuilding that locally
<efraim>divya: I haven't been able to build it, cargo complains about missing crates when I try to build it with the inputs listed in Cargo.toml
<cbaines>indeed, there's probably something that rust uses that uses meson
<divya>efraim: Ah, okay. I'll send another patch once I get back to my system.
<efraim>divya: thanks
<efraim>all of rust or just the last rust? because the bootstrap rusts don't use a lot of inputs
<csantosb>Do we have packages using #:install-plan with several outputs ?
<efraim>csantosb: perhaps take a look at gnu/packages/browser-extensions
<efraim>docbook also has some interesting install plans that use a '#$dest-dir'
<apteryx`>efraim: all of it
<efraim>ugh that's unfortunate
<apteryx`>but feel free to check, perhaps I was touching something more and my memory is off
<apteryx`>talking about rust, I'm packaging libcst, and I'm getting such an error in tests: read_dir: Os { code: 2, kind: NotFound, message: "No such file or directory" }
<efraim>sounds like a file isn't included
<apteryx`>backtrace: https://paste.debian.net/1337089/
<efraim>I think it's the same error for a socket for networking not existing
<csantosb>efraim: got it, thanks !
<apteryx`>efraim: OK! I guess it comes from this file: https://github.com/Instagram/LibCST/blob/08da127e5434eb42bae5dc71c03363d6a281d4bc/libcst/tests/test_roundtrip.py#L13C1-L13C86
<efraim>yep, looks like ../../../native/libcst/tests/fixtures
<efraim>for the --skip=roundtrip_fixtures I'd add ;; Not all files included
<apteryx`>efraim: indeed the native topdir is missing
<apteryx`>I'll report upstream
<janneke>divya: any idea how i get black-on-white with alacritty?
<divya>janneke: You can just try one of the several color schemes: https://github.com/alacritty/alacritty-theme
<efraim>sound better than my suggestion of changing colors.primary to add 'foreground = "#ffffff"' 'background = "#000000"'
<efraim>oh, it looks like "black" and "white" are accepted as colors
<janneke>divya: yeah, ive been searching for quite some time now and thought it ought to be easier than downloading and trying themes...
<janneke>ACTION tries efraim's suggestion
<efraim>man alacritty.5 is what I was looking at
<efraim>I also might have my colors backwards
<janneke>hehe
<andreas-e>efraim: Have you got a time frame for the rust branch, or should I sneek in gsl-upgrade before rust? I think the gsl update poses no problems, it should just be built before merging.
<efraim>andreas-e: as soon as QA says its good, so hopefully by the weekend
<andreas-e>Great, then I will wait!
<janneke>efraim: "failed to parse rgb color white; expected hex color like #ff00ff
<janneke>dude
<divya>Yeah, you most likely can define your colors in [colors] of your TOML
<efraim>my alacritty.toml is super simple https://bpa.st/RPRQ
<divya>efraim: Living the suckless life in alacritty I see
<efraim>I just wanted the text a little smaller and to not have to fight TERM variables :)
<apteryx`>back when I was using xterm I had XTerm*faceName: Hack:size=11:antialias=true in .Xresources
<janneke>ACTION now uses http://paste.debian.net/plain/1337091
<janneke>it seems it doesn't take a DPI setting and you have to calculate the scale setting from the dpi; oh well
<efraim>6.6‽ how good is your eyesight?
<janneke>hehe, it's just what i came up when using Xft.dpi: 225
<janneke>but that's why i want to be able to scale, if my eyes get tired late at night
<divya>Is anyone else's system reconfigure breaking because of a build error from umockdev? No idea.
<janneke>ACTION pulled to cd26d76fed instead of HEAD
<janneke>to avoid mesa-induced rebuilds
<efraim>great, upgrading vim isn't going to be straight forward, the tests crashed on lack of curl
<efraim>I'll look throught the ~280 commits later, I already have the checkout from the previous times
<apteryx`>how often does mumi index the rsync retrieved GNU debbugs data?
<efraim>15 minutes?
<apteryx`>do you know where that is defined?
<apteryx`>it kind of defeats 'mumi send-email', which tries for 15 minutes to get the new id
<apteryx`>(it always fail)
<aemogie>divya: same issue here
<divya>aemogie: :/ The mess continues.
<csantosb>Output of `guix search git` gives "outputs: + send-email: [missing description]"
<csantosb>Where exactly the missing description is supposed to appear in the package ? 🤔
<gabber>csantosb: that is only for the specific outputs of that package. you see the description of the "whole" git package on bottom
<csantosb>Yes, exactly. When one develop a custom channel, it's great to have this info along with the given output
<csantosb>"out" is clear, but "sim" or "dev" ?
<jaadu>I am trying to figure out guix home. How do I think about normal files in a managed home directory.
<jaadu>*?
<gabber>jaadu: as read-only ;)
<jaadu>So what happens with my documents? It doesn't seem useful if I cant edit them anymore
<jaadu>gabber: I thought so too, but I created a guix home container and I could create directories just fine.
<divya>jaadu: You edit them through Guile in your configuration file
<gabber>well, you may not want to have *documents* in your guix home config, but you might want configuration files, so guix keeps track of your changes and enables you to roll back to previous generations
<jaadu>But how do guix separate between my own *documents* and config files?
<gabber>through your guix home configuration ;)
<gabber>have you read the relevant passages in the manual?
<jaadu>gabber: How? How does guix knows if the particular .vimrc is a config file or vacation photos?
<jaadu>For example
<divya>jaadu: There are procedures in Guix that are specific to particular types of file.
<gabber>jaadu: have a look at home-dotfiles-service-type
<jaadu>I don't understand. What happens with my vacation photos? Do I have to mount it on a separate partition?
<divya>jaadu: Guix only touches the specific files you want it to.
<jaadu>divya: And otherwise it compares it to the previous generation, when you update your home-config?
<divya>Yep
<jaadu>I guess that makes sense
<jaadu>Currently I have a repo with dotfiles that I symlink to the appropriate places with a script.
<jaadu>I feel like I am halfway towards a guix-home config already.
<apteryx`>efraim: looking at 'sudo grep 'Sleeping for' /var/log/mumi.worker.log' on berlin, it seems the update-state! proc of mumi runs roughly every 30 *minutes*
<apteryx`>while the code suggests this should happen every 30 seconds
<apteryx`>something seems off
<aemogie>divya: going through the recent commits, might be the meson update (which i think umockdev uses). going to try building before it, ill give an update on how it goes
<jaadu>But I haven't figured out how to create a guix home service to create the symlinks I need.
<apteryx`>csantosb: I'm not sure if that's currently possible. lilyp perhaps knows
<divya>aemogie: Please do.
<apteryx`>(to give a custom description to extra labels on per-package basis -- I don't think that's possible, and I'm not sure it's a good idea)
<apteryx`>i'd prefer a very restraint set of output names that are defined at one place and described there
<apteryx`>via some enum say, that'd be checked at build time
<jaadu>There is a symlink procedure but I think it only expects to create symlinks to the store.
<apteryx>rekado: hi! it seems mumi is not progressing much with the indexing of emails according to /var/log/mumi.worker.log on berlin: https://paste.debian.net/1337103/
<gabber>ACTION is about to try installing GNU+Hurd on a T410
<apteryx>rekado: a bug in mumi it seems: 'mumi worker' calls (update-state! #:loop? #t), but when this loops to itself it looses the #:loop? argument, thus it effectively doesn't loop.
<jaadu>Why hurd?
<gabber>jaadu: because i can (:
<apteryx>it's an elegant design compared to monolithic kernels
<gabber>jaadu: also, i can finally show off why it is essential to keep kinda-old-but-not-vintage-yet hardware laying around
<apteryx>rekado: I think this should fix it: https://paste.debian.net/1337104/
<divya>gabber: Same. I'm keeping my 16 year old i3.
<csantosb>apteryx: I think we need to consider custom, non standard outputs in personal channels
<csantosb>I mean, guix is not limited to software. It handles dependencies in other fields.
<oriansj>has anyone packaged https://git.sr.ht/~bptato/chawan yet?
<jaadu>gabber: I approve of your commitment
<apteryx>rekado: confirmed the fix locally; will tag, update package in guix, then update on berlin
<apteryx>csantosb: perhaps the enum mechanism I imagine could be extended by channels
<apteryx>this way they could extend/standardize in their own space
<apteryx>having a soup of whatever labels would hurt scripting
<apteryx>say you want to install all the doc of all the packages in your manifest; you want the doc output named 'doc', not 'doc-html', 'doc-legacy', or worse.
<jaadu>This is as far as I get: https://paste.debian.net/1337105
<futurile>oriansj: toys says nope: https://toys.whereis.social/?search=chawan - all yours ;-)
<divya>the meson issue broke nongnu too I guess and now I'm left with the unsupported manifest format error. This has been a wonderful week :)
<apteryx>civodul: how long of a wait time is there if any before a shepherd services restarts a process that exited cleanly?
<apteryx>(hello!)
<aemogie>divya: cd26d76fedb7ab13ad91bd5dcfce119892b8e62e works
<oriansj>futurile: thanks for the link, I'll add it to my notes
<apteryx>I'm seeing 30 minutes on the current mumi-worker service on berlin, which seems abnormal.
<apteryx>the service runs for 30 s then exit currently, because of a bug
<apteryx>I'd expect shepherd to step in and resumes its operation promptly
<divya>aemogie: Thanks, let me try.
<gabber>i run into a kernel panic when running the installer
<divya>gabber: Welcome to my life for the last few days :)
<gabber>are you trying to install the hurd as well?
<civodul>apteryx: hi! there’s ‘respawn-delay’, which defaults to 0.1s
<civodul>“sudo herd log |grep mumi-worker” on berlin shows that it’s restarted immediately after exit
<divya>gabber: No, but trying to modify the linux-libre to make it work with my processor. And yes, upon janeke's suggestions I've begun messing around with hurd as well.
<civodul>(that mumi-worker exits every hour isn’t a good sign :-))
<gabber>janneke: what commit did you build the Guix installer with Hurd batteries included?
<civodul>mumi-worker seems to exit with zero, so maybe it’s on purpose?
<apteryx>civodul: indeed! it's just processing all that time
<apteryx>so one full email database indexing pass takes about 30 minutes
<apteryx>and it attempts this every 30 s
<apteryx>(currently)
<apteryx>that's bad. there's also guile warnings about heap size
<apteryx>in /var/log/mumi.worker.log: 2024-11-27 16:24:40 GC Warning: Repeated allocation of very large block (appr. size 32768 KiB): 2024-11-27 16:24:40 May lead to memory leak and poor performance
<apteryx>that's one reason why 'mumi send-email' never seems to pick up a newly created ID
<civodul>oooh
<apteryx>the remote database it queries is too slowly refreshed
<civodul>i always wondered why
<apteryx>raw data is currently synced over rsync every 5 minutes
<apteryx>that works
<apteryx>but 'mumi worker' is struggling to index it
<apteryx>rekado: update-state! has some code path to do a partial update (#:full? #f), but that's not currently used; would it help here?
<apteryx>"When FULL? is #T process also archived issues."; I guess we do not want to do this
<apteryx>every 30 s
<hako>Is lsof broken on master? I got test failure on my machine: <https://paste.sr.ht/~hako/7cb3369c1c4881c1f7061344f95308793b3c1d64>.
<apteryx>known for years it seems: https://lists.nongnu.org/archive/html/bug-guix/2020-11/msg00579.html
<apteryx>we haven't yet skipped it?
<apteryx>same issue: https://github.com/lsof-org/lsof/issues/152
<hako>🤔
<apteryx>btrfs may have a patch for it, based on this old work: https://patchwork.kernel.org/project/linux-btrfs/patch/20130710174545.GS32502@wotan.suse.de/; worth checking
<apteryx>I meant, Suse
<apteryx>hm, not sure that one is relevant.
<apteryx>hako: I don't really see anything there: https://build.opensuse.org/package/show/openSUSE:Factory/lsof
<apteryx>I'd skip it, until it's properly resolved upstream
<hako>It was disabled, but lost in fb59f4f392635f459ed199a3d291dbdb7c437aae it seems.
<apteryx>oops.
<apteryx>I bet they assumed it was resolved after the patch didn't apply cleanly and the tests passed for them
<janneke>Git error: cannot locate remote-tracking branch 'origin/keyring'
<janneke>what do i do to fix this again?
<janneke>ACTION already did: git branch keyring origin/keyring
<janneke>ACTION did a fresh git clone gnu:guix
<ekaitz>classic tracking branch issue hehe i had that yesterday
<janneke>ACTION runs guix git authenticate
<janneke>hmm, that's not it
<janneke>ekaitz: how did you resolve it?
<efraim>janneke: you might have to checkout the branch, i'm not sure if 'git branch' would take care of it
<cbaines>janneke, I think the git fetch origin keyring:keyring from here should help https://guix.gnu.org/manual/devel/en/html_node/Building-from-Git.html
<cbaines>(I ran in to this recently)
<janneke>cbaines: yeah, i did that
<janneke>redoing it just silently returns 0
<janneke>ekaitz: checking out the branch (in a worktree) doesn't help
<old>is there a reason why package pkconf does not make a symlink of the form: pkg-config -> pkgconf ?
<old>s/pkconf/pkgconf
<old>would be nice to have, since this would make it easier to simply change pkg-config to pkgconf for various packages no?
<divya>The weirdest thing is that when I do a fresh install of guix, it works with linux-libre on my cpu just fine it breaks after modifications
<janneke>ACTION necrobumps https://issues.guix.gnu.org/57229
<attila_lendvai>is there a way to initiate QA to recheck an issue? it says there's no patch there, which is not true (https://issues.guix.gnu.org/73148)
<janneke>ACTION found a workaround https://issues.guix.gnu.org/57229#3
<janneke>re: cannot locate remote-tracking branch
<vagrantc>huh. just starting to realize ... guix has not released since the last Debian release ... and Debian is probably going to start the freeze process soonish... might see guix 1.4.x in debian for two consecutive releases!
<the_tubular>Yeah, since 1.3.0 guix releases seems to have slow down
<vagrantc>"I remember when you were a wee little thing. Now look at you!"
<janneke>vagrantc: hehe
<lilyp>csantosb: you put the output message as a property, see e.g. the git package
<rekado>is anyone in charge of coordinating the next release?
<vagrantc>i vaguely recall a brief burst of activity, but i think there were problems with infrastructure around the same time and maybe lost traction?
<vagrantc>looks like a thread in early september...
<vagrantc>previous thread in may...
<rynn>If I have an issue with the build of a package failing, what would be the starting point for figuring out the culprit?
<ieure>rynn, The build output and/or logs.
<ieure>Path to the build log should be in the output of the failing build.
<rynn>I checked the log dump, but only thing I'm really spotting is an uncaught exception: %exception #<&invoke-error program: "ninja" arguments: ("-j" "16") exit-status: 1 term-signal: #f stop-signal: #f>
<rynn>While building umockdev
<ieure>rynn, Look further up, that's the build-system telling you the command that failed, the output from the command is what you want.
<efraim>I just fixed umockdev
<rynn>Do you mean there was something fubar in the package def efraim?
<efraim>there was a change in meson that exposed warnings from code generated by vala, and there was a CFLAG in umockdev that would error on one of those types of warnings
<efraim> https://github.com/martinpitt/umockdev/issues/228
<gabber>pulling master results in: "guix git: error: unknown introductory commit and signer". what do?
<ieure>gabber, Did you recently `guix system reconfigure'?
<ieure>gabber, Assuming this is master of Guix itself. I think the only way you'd get that is if the Guix daemon was misconfigured & doesn't know about the signing keys for the channel.
<gabber>not *that* recently. also, i'm talking about `git pull` in the repository
<ieure>I do not see how you could possibly get that error from a `git pull', since.
<ieure>*since the error comes from Guix.
<ieure>Git hooks in that repo maybe?
<gabber>ieure: me neither. perhaps through some funky stuff in guix git config?
<gabber>the repo is guix itself
<vagrantc>pulled from a (local?) branch without updating the corresponding keyring branch?
<gabber>vagrantc: now this sounds plausible!
<gabber>how do i do that/what do i have to do?
<vagrantc>well, my workflow is guix pull --url=/path/to/my/local/guix/checkout --branch=master ...
<vagrantc>but if you are not using a similar workflow, that is probably not the issue
<gabber>ACTION is confused
<gabber>so should i pull/reconfigure my system to make it recognize the newly added keys?
<vagrantc>what exactly are you doing? it is unclear from what i have read so far
<vagrantc>gabber: are you just doing a bare "guix pull" with no arguments? do you have any channels configured?
<gabber>no. i am trying to update my local guix repo checkout (on branch master) which raises this error
<ieure>gabber, Anything obvious in .git/hooks/ that'd be triggering Guix when you `git pull'?
<ieure>pre-receive is probably the most likely.
<ieure>Ohhh okay it's gotta be .git/hooks/post-merge, which triggers `guix git authenticate'
<gabber>ieure: i didn't configure any hooks - like at all. ever
<ieure>gabber, guess: ekaitz pushed commits, but your guix doesn't have their key yet.
<gabber>so i pull/reconfigure my system then :)
<ieure>Yeah, I think that's likely to fix things.
<gabber>now that efraim fixed the master branch i should be safe ;)
<vagrantc>guix doesn't have any keys, it just uses the keyring branch "next" to the master branch...
<vagrantc>and the .guix-authorizations files...
<ieure>vagrantc, Hmmm. I've never checked out the keyring branch on my clone, how does `guix git authenticate' get at its contents?
<gabber>guix pull fails bc of channel `guix-past'
<ieure>gabber, In your clone, do you have the keyring branch checked out?
<ieure>Wondering if you need to `git remote update' to get the commit to the keyring branch which `guix git authenticate' needs to check master.
<gabber>ieure: i had some old checkout of that branch locally
<csantosb>lilyp: not sure to follow, git package gives precisely no description in its outputs
<ieure>gabber, Suggest switching to it and pulling, or alternately deleting the branch.
<lilyp>indeed
<gabber>ieure: thanks!
<gabber>where is that guix-past channel coming from (i see it through `guix describe') if not from /etc/guix/channels.scm or ~/.config/guix/channels.scm ?
<gabber>ACTION getting more and more confused
<lilyp>there are two examples actually using it: remind and openscenegraph
<csantosb>I see now what you mean, thanks !
<ieure>gabber, Could be in your operating-system config. You can specify system-wide channels with the channels field of guix-configuration.
<ieure>gabber, That's not a default channel, so something you did has to have been how it got added. Agreed that it is curious, though.
<gabber>huh. even `guix pull --channels=channels/file/without/guix-past.scm` fails - on channel guix past
<gabber>is it particularly stupid/dangerous to `./pre-inst-env guix pull' from my checkout?
<vagrantc>ieure: well, normally guix pull defaults to using the upstream guix.git master branch, which also has the keyring branch ... if you pull from a local clone of guix.git master branch you need the keyring branch too.
<x8dcc>Hello, I have these lines inside '(operating-system (services ...))', but the keyboard layout is not being set correctly. The value of the 'keyboard-layout' variable is fine, since it's being set in other places like TTYs or the bootloader. https://bpa.st/OIBQ
<gabber>how did ~/.config/guix/channels.scm end up in the store?
<vagrantc>not sure how this plays with non-default channels ... or maybe something changed in the workflow since tthat was required some years ago
<gabber>i'll wrap it up for today
<gabber>ACTION goes afk. *confused gabber music*
<ieure>gabber, If you use Guix Home and specify channels there, ~/.config/guix/channels.scm is a link into the store.
<gabber>*gabber music fading out* this makes sense. but even there i have not configured guix-past
<ieure>Yes, quite a mystery.
<x8dcc>Apart from my previous question, does anyone use 'startx-command-service-type' successfully? https://guix.gnu.org/manual/devel/en/html_node/X-Window.html#index-startx_002dcommand_002dservice_002dtype
<ieure>I've never tried that. I wonder why startx is so much harder to use on Guix? Does it need suid and not get installed like that?
<x8dcc>No clue, but I had to make my own script for it, following some link someone here sent
<x8dcc>It works fine, but I rather use the normal 'startx' command
<x8dcc>The problem when using 'startx' in my machine is that there is no 'X' binary under '/gnu/store/...-xinit-1.4.2/bin/', there is only 'startx' and 'xinit'
<janneke>x8dcc: something like: startx -- $(guix build xorg-server)/bin/Xorg
<janneke>(untested)
<janneke>but having it fixed in the xinit package would be nice
<x8dcc>I can try that in a bit (or perhaps tomorrow), but I don't really understand how that has remained broken (if it really is broken)
<janneke>x8dcc: in the old days we'd all startx
<x8dcc>Every Guix user (that uses Xorg) had to make his own 'startx' script?
<x8dcc>janneke: Hmm... What do you mean by that?
<janneke>x8dcc: i guess that nowadays all use lightdm or even gdm
<aldum>iirc it was an .xinitrc, and startx would read that
<janneke>aldum: yeah, but our startx script is broken
<aldum>also, the X[org] binary is usually not in the xinit package
<x8dcc>Yeah, I have been using '.xinitrc' for ages
<x8dcc>aldum: Do you have any idea on how to fix that? I mean, the problem is that it got removed from the xinit package, or did the old script overcome that?
<ekaitz>janneke: did you fix the branch issue?
<ekaitz>i was afk
<ekaitz>janneke: i fixed it with a `git checkout keyring`
<janneke>ekaitz: i found a workaround: https://issues.guix.gnu.org/57229#3
<janneke>ekaitz: yeah, already did that -- my case was a bit tricksier
<ekaitz>oh i think i see
<ekaitz>just tracking the branch with the remote was not enough
<janneke>anyway, thanks for checking
<ekaitz>no problem
<aldum>x8dcc: (disclaimer: not guix-specific, it very well might be different here) my point is, it should be in xorg-server or something like that, xinit is a separate package
<janneke>ekaitz: hey and congrats with your commit acces!
<ekaitz>janneke: thank you for supporting it
<ekaitz>:)
<janneke>:)
<janneke>you're welcome
<ekaitz>now some people is trying to make me become a guile commiter but let's relax for the moment
<aldum>x8dcc: i.e. /gnu/store/qanzv0plcsa7k61ky6q2gja4n0swfm94-xorg-server-21.1.12/bin/Xorg
<shrini-weechat>hello all
<shrini-weechat> https://logs.guix.gnu.org/
<shrini-weechat>how this logs and web interface is achived/
<shrini-weechat>?
<shrini-weechat>what is the logbot used here?
<shrini-weechat>logs are pretty
<ekaitz>shrini-weechat: gogles it think it's called, let me link you
<ekaitz>shrini-weechat: https://git.savannah.gnu.org/cgit/guix/maintenance.git/tree/hydra/goggles.scm
<shrini-weechat>thanks
<shrini-weechat>can we install the bot in a non guix server?
<ekaitz>you probably can, it's just a guile program
<ekaitz>btw rekado https://git.savannah.gnu.org/cgit/guix/maintenance.git/tree/hydra/goggles.scm#n5 <-- your email goes to civodul 's inbox, interesting tactic
<shrini-weechat>thanks ekaitz
<shrini-weechat>will check on how to use it
<shrini-weechat>is there any other documentation on how to install and configure it?
<shrini-weechat>on a ubuntu or debian server?
<shrini-weechat>I like to take logs for few IRC channels and host the logs
<ekaitz>let me take a deeper look
<x8dcc>aldum: Replying to what you said about the Xorg binary: I also agree that it should be moved to a different package, but I think 'xinit' should either depend on it, or just locate the binary on the (potentially) correct path
<ekaitz>shrini-weechat: you have two files in that project you could read: goggles-bot and goggles
<ekaitz>shrini-weechat: one is the website and the other one is the logger
<ekaitz>the -bot is the logger, it's pretty easy to run
<ekaitz>and the other is the website, just defined in that file
<ekaitz>shrini-weechat: if I were you, I'd try to use those two files independently
<x8dcc>aldum: It seems that the 'xinit' package (inside 'gnu/packages/xorg.scm') just downloads 'xinit-VER.tar.xz' without patching or modifying anything
<shrini-weechat>ok
<shrini-weechat>i am all new the world of guix, guile, schema
<shrini-weechat>scheme
<ekaitz>shrini-weechat: do you have experience in any other programming language?
<shrini-weechat>python
<shrini-weechat>and ruby
<aldum>interesting, it's not dependent on xorg-server on Arch either
<shrini-weechat>the bot code seems like elisp code, that I see on the packages of emacs
<shrini-weechat>if possible, write a tutorial on how to setup this logbot
<shrini-weechat>for a non programmer
<shrini-weechat>it will help many others to setup the logging in their channels
<ekaitz>shrini-weechat: scheme is similar to elisp, but it's also very similar to python
<ekaitz>you can learn scheme's fundamentals in this tutorial series: https://diode.zone/w/bi4GDSW54SXBAcCYnebKz7 (7 videos)
<ekaitz>about guile specifically, the manual is a good resource
<ekaitz>and about the logger... there are many IRC loggers, this is only a simple example of one. In fact, you could build your own "easily"
<shrini-weechat>ok
<ekaitz>goggles-bot: hi
<ekaitz>lol
<janneke>amazing alacritty doesn't have bash/emacs keybindings? -- by default
<janneke>oh wait, line mode, my bad
<futurile>does anyone specifically handle security or X?
<futurile> https://issues.guix.gnu.org/issue/74248#3 seems to be a CVE patch
<ekaitz>janneke: the keybindings are from the shell, not the emulator
<ekaitz>(well, the ones you are looking for)
<jonsger>futurile: there is a security team
<futurile>jonsger: guess no-one notified them
<janneke>ekaitz: yeah
<ekaitz>janneke: btw I use alacritty since a couple of years ago and I don't dislike it. It's fast.
<futurile>OK, I've emailed the list, hopefully someone will see it and check if it's a real security issue. It looks like they provided a patch, so at first glance seems like a real thing
<Rutherther>why hasn't this simple patch been merged? https://issues.guix.gnu.org/70672 also should be done for python-next
<divya>Has there been any progress in guix being able to mount NTFS file-systems on boot?
<divya> I have to manually do that everytime, if there's been no progress maybe one can use a simple-service to employ the mount.ntfs-3g command?
<divya>Saw this issue, almost 3 years ago: https://lists.gnu.org/archive/html/help-guix/2021-10/msg00039.html
<ekaitz>Rutherther: perhaps because it affects many packages?
<Rutherther>ekaitz: yeah, but why couldn't it go to python-team?
<ekaitz>it could