IRC channel logs

<neonio>is anyone here running windows games on guix? if so, how? both the Bottles and Lutris flatpaks have strange issues for me

<neonio>Lutris cannot find the certificates to download runners, and bottles is having some kind of issue locating graphics drivers(?)

<neonio>I'm reluctant to bring these issues to these projects' repos 'cos I'm not sure if they're just esoteric Guix issues

<bdunahu>I haven't used flatpak, but if it's really doing sandboxing/containerization where the sandbox contains the full runtime, then the programs probably don't interface much with guix or its unconventional filesystem layout?

<efraim>neonio: for some flatpak packages I used flatseal to add "SSL_CERT_DIR=/etc/ssl/certs" to the environment variables passed through

<folaht>Hey there, I've got some issues. I can no longer do a 'guix pull'

<apteryx>what's the error?

<folaht>error: use-modules: unbound variable

<apteryx>do you see the name of the unbound variable?

<folaht>hint: Did you forget `(use-modules (guile))' or `#:use-module (guile)'?

<folaht>ice-9/eval.scm:223:20: In procedure proc:

<folaht>No name that I'm aware of

<apteryx>weird

<apteryx>if 'guix pull' is broken than way for everyone it'll be fixed soon

<efraim>do you have other channels defined?

<janneke>folaht: could it be your channels file?

<folaht>efraim, yes. How do I undefine my other channels?

<janneke>ACTION needs to use --unsafe-channel-evaluation, as they had (introduction (@@ (guix channels) %guix-channel-introduction))

<janneke>all over the place; but it seems that's no longer allowed :-(

<efraim>you could just comment them out, but then you wouldn't get to use the packages/services from those channels

<folaht>There's a lot not working right now on my end and I don't know if this is my doing or if this is an issue with guix anymore.

<folaht>I commented out the other channel that I have in my channel list, so it now only has %default-channels. 'guix pull' give me the same error though.

<apteryx>folaht: to disregard temporarily your channels, you can also 'guix pull -q', IIRC

<folaht>apteryx, hey that works! Or at least it tries to do something.

<folaht>Alrgiht, I have another issue. When I try to do 'guix upgrade' I also get an error

<apteryx>mv ~/.config/guix/channels.scm{,.bak} and try again

<folaht>guix upgrade: error: canonicalize-path: No such file or directory: "xkeyboard-config-yr.patch"

<apteryx>I remember you were experimenting with patching xkeyboard-config last time; maybe you have some clean-ups to do in your config?

<folaht>Yeah, but the xkeyboard-config-yr.patch is commented out.

<folaht>So even when it's commented out and having done a 'sudo guix reconfigure /etc/config.scm' I still get this error.

<yarl>home-reconfigure?

<yarl>guix package -r ?

<yarl>upgrade is for packages installed by `guix package`

<yarl>(it's an alias)

<folaht>Ahhh... thanks yarl. Unistalling xkeyboard-config seemed to do the trick.

<folaht>And hurray! Kitty finally got its upgrade!!

<apteryx>glad you got that sorted

<apteryx>are there known ipv6 issues with make-systemd-constructor and containerized processes?

<folaht>Yeah, I've been confused about that for days now. I kept thinking it had something to with my system configuration.

<apteryx>ipv6 on my service mysteriously stopped working after using least-authority-wrapper on it

<futurile>morning all

<efraim>o/

<apteryx>futurile: o/

<apteryx>do we have an example that uses this combo: make-systemd-constructor + least-authority-wrapper + ipv6 (e.g. localhost in a test) ? It seems broken.

<user_oreloznog>\o

<civodul>o/

<sneek>Welcome back civodul, you have 1 message!

<sneek>civodul, apteryx says: hi! I guess it's expected that we can't 'setenv' in the shepherd process (e.g. in a one-shot service lambda?)

<civodul>apteryx: hey! you can call ‘setenv’ just fine i guess? though i wouldn’t recommend changing the environment of the shepherd pross

<civodul>*process

<apteryx>I can call it, but then an application I run with invoke does not "see" the value set

<apteryx>I was trying to (setenv "HOME" "/var/lib/forgejo") for example, and then running forgejo complained it could not find HOME (with its code doing getenv HOME)

<efraim>wrap the whole thing with begin? or (invoke "FOO=bar" "baz" "--flag")

<apteryx>civodul: it was in a one-shot that looked like this: https://codeberg.org/guixotic/guix/src/branch/add-forgejo-package-and-service/gnu/services/forgejo.scm#L551 (current code uses a workaround)

<apteryx>efraim: does invoke accept environments as first argument? IIRC it wants an executable directly

<apteryx>you could go through "sh" "-c" "ENV=... the-command" perhaps, or "env", but that's not super elegant

<efraim>good question, but you can always add more redirection! (invoke "sh" "-c" ...

<identity>‹system› does, but it has the same problems as ‹system*›

<efraim>or zero? system

<efraim> https://gitlab.com/Efraim/shepherd-services I've had some ugly ones in the past that I used because they Just Worked™

<apteryx>maybe the shepherd should protect itself against environment mutations by forking before evaluating one-shots? I think in this case the setenv would probably work (I'm guessing its environment gets reset in the concurrent fibers execution somewhere, which is what would break my previous use case?)

<efraim>would that protect against my R-as-PID1 r-shiny service?

<efraim>ok, I still can't use 'guix style' in gnu/packages/rust-sources.scm, but I can now build the packages there with -e

<apteryx>civodul: in case you're in a shepherd hacking mode/mood, I'm having an issue with make-systemd-constructor and least-authority-wrapper; I've dumped my debugging thus far in https://codeberg.org/shepherd/shepherd/issues/122

<apteryx>in the past I've seen the same combo break tls connections, maybe it's related

<egnuun>Hello, Guixers,

<egnuun>I have a quick question. Working with Firefox in Guix shell works… I have no audio, but that is not my most pressing problem today. ;-)

<egnuun>When I want to "Ctrl+S" an image or apparently when I even just want to move some bookmarks with my mouse. It always crashes. Do you know why?

<tusharhero-xmpp> https://commons.wikimedia.org/wiki/File:GNU_Guix_logo.svg

<egnuun>The error message is something like this:

<egnuun>Failed to create DBus proxy for org.a11y.Bus: Cannot spawn a message bus without a machine-id: Unable to load /gnu/store/[…]/var/lib/dbus/machine-id or /etc/machine-id: Failed to open file “/gnu/store/[…]-glib-2.86.0/var/lib/dbus/machine-id”: No such file or directory

<tusharhero-xmpp>what did this logo used to mean?

<futurile>egnuun: are you making dbus available within 'guix shell'?

<egnuun>How would I do that?

<futurile>egnuun: do you know what dbus is?

<egnuun>I do know, what dbus is.

<egnuun>Ah, yes. I used the example command from the guix docs:

<egnuun>guix shell --container --network --no-cwd ungoogled-chromium \

<egnuun> --preserve='^XAUTHORITY$' --expose="${XAUTHORITY}" \

<egnuun> --preserve='^DISPLAY$' -- chromium

<egnuun>But without the "--no-cwd"

<futurile>egnuun: OK, you'll find this is a general pattern for a lot of GUI apps because 'guix shell' doesn't provide a connection by default. You probably need stuff like --preserve='^DBUS_' --expose=/var/run/dbus

<futurile>egnuun: I did a blog post on it where I used Firefox (https://www.futurile.net/2023/04/29/guix-shell-virtual-environments-containers/), it's probably a bit out of date at this point, but might help

<egnuun>Ah, thanks, futurile. I just found this git repo:

<egnuun> https://github.com/nuthub/guix-shell-examples

<egnuun>Which has the same solution.

<futurile>nice

<egnuun>I guess my question is, why not include these in the general docs?

<futurile>normally the answwer is "because no-one has submitted a PR". But, often the manual trys to be comprehensive about how something works, but isn't comprehensive about "how to do X". The 'howto' part is supposed to be in the Cookbook. It's not a great division of labour honestly.

<m4xxed>Heya, I have tried forking the guix/guix repo on codeberg multiple times in the last 2 weeks now, only to always get a 504 error (the server didn

<m4xxed>'t respond in time)

<m4xxed>anybody else having those issues? is this common? or is there something wrong with my account maybe?

<sham1>It tends to happen often, yes. It's probably because of how large the guix repo is

<identity>m4xxed: you are probably better using AGit-Flow anyway; see <https://forgejo.org/docs/latest/user/agit-support/>

<m4xxed>identity thanks, I will look into that :)

<yarl>Hi civodul, are you busy?

<futurile>I'm very unsure if "discussing" GCD's via Issue is a good idea

<futurile>ACTION looks at the AI/LLM issue gathering a lot of comments in an unthreaded format

<futurile>but maybe I just need threading because my working memory is getting smaller these days heh

<civodul>futurile: yeah, it’s not very convenient

<civodul>anyone knows what’s up with KeepassXC? “WARNING: You are using an unstable build of KeePassXC.”

<bjc>when do you see that and where?

<bjc>i haven't updated for about a week, but my version (2.7.12) doesn't seem to spit out a warning that i can see

<civodul>there’s a big yellow banner at the top of the GUI

<identity>it seems it complains because it is a snapshot build and not a release build

<civodul>it’s 2.7.12

<civodul>what do you mean, “snapshot build”?

<identity>well, it *thinks* it is a snapshot build

<civodul>ah, yes

<civodul>yeah that’s weird

<bjc>if we're both running 2.7.12, why isn't mine complaining?

<identity>CMakeLists.txt has (the equivalent of) if(KEEPASSXC_BUILD_TYPE STREQUAL "Snapshot") set(KEEPASSXC_BUILD_TYPE_SNAPSHOT ON) endif(), and MainWindow.cpp has #if defined(KEEPASSXC_BUILD_TYPE_SNAPSHOT) […]showMessage("WARNING: You are using an unstable build of KeePassXC.\n"[…])

<identity>which is the only place where «unstable build» in the source (disregarding translation files)

<identity>appears in the source

<bjc>on master, the sha/version hasn't changed since the mid-march

<bjc>so we should all be running the same version, and yet i don't see any warnings

<bjc>and 2.7.12 is an official release, not a snapshot

<Maurius>Hello after recent guix upgrade some dependencies of keepassxc were updated and after that I'm getting this: https://limewire.com/d/E5ZH7#7qvRlpXECx (before it was without the warning) is it normal ?

<identity>Maurius: i cannot open the link, but let me guess: «WARNING: You are using an unstable build of KeePassXC. [&c.]»?

<Maurius>yes exactly I've checked logs from this channels and it seems that someone else was also complaining about it today

<yelninei>keepassxc got updated to build from git directly, maybe the tarball has some extra magic included?

<identity>probably

<civodul>ah, that would explain it

<yarl>Hi civodul, do you have few minutes?

<yarl>Regarding #8681

<bjc>i haven't bees able to build my system with (service cups-service-type) for some weeks now due to inability to compile libpaper for some reason

<yarl>I guess you want proper lists serialized as (list 'value SOMETAG a b c) and improper lists as (list 'value SOMETAG a b . c)

<bjc>if i ‘guix build libpaper’ by hand, it's fine

<yarl>and SOMETAG serves to tag non self quoting values e.g.:

<yarl>(let ((sometag (gensym))) (list 'value sometag 1 2 (list sometag (object->address value) (object->string value))))

<yarl>Is that what you meant?

<yarl>more (list 'value SOMETAG (a b c)) I guess

<yarl>Please tell sneek if I'm on track if you may, I got to go.

<apteryx>yikes, can't fetch from codeberg

<apteryx>seems transient errors

<yelninei>mmh appearently misc-world-rebuild is a full rebootstrap due to touching (guix build utils)

<civodul>sneek: later tell yarl sorry for missing you earlier! i would serialize pairs (be they proper or improper lists) as (list 'value TAG the-pair)

<sneek>Okay.

<futurile>bjc: you build it by hand, but it fails when you build with a system update? Did you get an error message?

<bjc>futurile: it only occurs when building cups. i have no idea why

<bjc>if i build it by hand there's no problem. in fact, there's a substitute

<bjc>so i have no idea what's going on

<bjc>i assume there are shenanigans afoot in the service definitition, since the cups package builds fine by itself, too

<yelninei>bjc: Maybe its the 32bit version for wine or something?

<bjc>yelninei: i'm not using wine. i didn't ever realize cups had hooks for that

<bjc>it's weird that it's seemingly just me. no other reports in the issue tracker that i see, and i've been having this problem for weeks

<bjc>i don't know when it started, though. i only tried to add the cups-service-type a couple weeks ago, and it's never worked

<folaht>I have for: "sudo guix system vm /etc/config.scm" the error: "patches: unbound variable, hint: Did you forget a `use-modules' form?"

<folaht>How do I know which use-modules is missing?

<yelninei>bjc: can you take a look at the log and see what is failing?

<bjc>i can and have! https://bpa.st/C2HA

<bjc>error is at the bottom

<bjc>i can't figure out where that error is coming from. i used -K and went looking in the source and came back empty-handed

<folaht>Oh wait, I can see now what I'm doing wrong. Never mind.

<bjc>i mean ‘make’ causes the error to show, as in that output. but i don't know where the ‘--help’ flag is coming from

<folaht>I forgot to remove brackets in the previous after commenting out the patches line.

<folaht>*previous line

<bjc>i don't know my way around help2man, though, so i might be misinterpreting that error

<futurile>bjc: but you've built it by hand AND you've used the substitute, so why is it trying to build it when you switch the service on

<bjc>yeah, i don't know

<bjc>it's very weird

<bjc>the service definition has to be doing something, but i don't see anything

<yelninei>bjc: --build=x86_64-unknown-linux-gnu" "--host=x86_64-pc-gnu, you are trying to cross compile libpaper and it fails to run the cross compiled binaries for help2man

<bjc>nono, this is regular linux

<bjc>i wanted to print some stuff from my pc =)

<yelninei>the libpaper you are building is cross compiled to 64bit hurd, did you add the service to the correct os?

<bjc>it… is…?

<bjc>wtf

<bjc>oh my god

<bjc>that was in my childhurd definition

<bjc>thank you!

<futurile>ah nice - always better when it's something simple - nice catch yelninei

<yelninei>definitly did not do that same mistake before

<bjc>haha

<yarl>o/

<sneek>yarl, you have 1 message!

<sneek>yarl, civodul says: sorry for missing you earlier! i would serialize pairs (be they proper or improper lists) as (list 'value TAG the-pair)

<yarl>right

<yarl>Seems less simpler but lighter than what I propose.

<tesseract>guys, could someone tell me if icecat was upgraded to 140.11.0 version in guix?

<tesseract>they released the source two days early. made me confused

<ieure>tesseract, You can look in the source yourself. https://codeberg.org/guix/guix/src/branch/master/gnu/packages/gnuzilla.scm#L608

<ieure>or `guix pull && guix show icecat'

<tesseract>so, it says "define %icecat-base-version "140.10.2""

<tesseract>ieure: i odn't use guix

<tesseract>don't*

<ieure>Okay... not sure why you care what version of Icecat Guix has, then.

<tesseract>ieure: what makes me confused, they fetch firefox source code from candidate section instead of "release" section from firefox archive.

<tesseract>i don't know why

<ieure>tesseract, What makes you think that? What difference would it make?

<ieure>Is "they" IceCat or Guix?

<yarl> https://ci.guix.gnu.org/search?query=icecat-140.11.0

<yarl>gnuzilla-updates

<tesseract>ieure: i think candidate isn't certain release, they can change the source anytime if something is wrong. on the other hand release came from candidate

<tesseract>but for example there can be 2 candidate

<ieure>tesseract, What evidence do you have to support this? I'm looking at the source, and it points to a release tarball: https://codeberg.org/guix/guix/src/branch/master/gnu/packages/gnuzilla.scm#L625

<tesseract>ieure: i am cloning from here https://gitweb.git.savannah.gnu.org/gitweb/?p=gnuzilla.git

<ieure>tesseract, That's the upstream source, not the Guix package.

<tesseract>i don't know this git stuff to be honest. i just found a guide and doing that

<tesseract>ieure: so, is it wrong to compile from that source?

<tesseract>damn. i don't know this stuff

<ieure>tesseract, It's not wrong, but it's not the Guix package, nor is it necessarily the precise source used for the Guix package. You're inferring the behavior of one system by looking at a completely different one.

<ieure>tesseract, Start here, you're not going to get far either developing or building FOSS stuff without a good understanding of Git: https://git-scm.com/book/en/v2

<tesseract>why is this exist then? what is the purpose of this: https://gitweb.git.savannah.gnu.org/gitweb/?p=gnuzilla.git

<tesseract>yarl: thanks

<ieure>tesseract, It's the source repository where Icecat is developed, which is also its purpose.

<ieure>Well, it's a web UI on top of the Git source repo, but, that is not a distinction you care about.

<tesseract>i mean, the icecat compiled from there works fine and has all addons etc

<ieure>So>?

<tesseract>so, should i get the source from there?

<tesseract>shouldn't i*

<graywolf>civodul: Hi, qq, what is reason behind set of trusted channels not being union of trusted-channels.scm and guix describe?

<ieure>tesseract, That's a subjective question I cannot answer for you. Why are you asking about any of this in #guix?

<tesseract>ieure: you are so hostile. i don't want to talk to you anymore. probably i will never join in here anymore. such hostility

<ieure>lol, okay

<folaht>I have kitty version 0.46.2 installed, yet my system still uses the old version.

<ieure>folaht, Foreign distro?

<folaht>No, I use guix system

<ieure>folaht, Are you launching it from a DE?

<folaht>Yes

<ieure>Log out and back in.

<folaht>I already fully rebooted.

<ieure>What DE?

<folaht>XFCE

<ieure>Is it running the kitty from your current profile generation? Maybe it cached the old generation's .desktop files.

<folaht>How can I tell?

<ieure>Launch it and look at the path to the binary, is it the same as what your current profile is linking to?

<ieure>ex. does `ls -l $(which kitty)' in a terminal point to the same path as the kitty launched from the menus?

<folaht>Yeah, it's pointing to the old kitty.

<folaht>How do I get it to point to the new one?

<ieure>No idea!

<folaht>lol

<folaht>I'll try guix home reconfigure then.

<ieure>Did you install kitty via some other mechanism than `guix home reconfigure'?

<folaht>I did a 'guix upgrade', but kitty is mentioned in my home-configuration scheme file.

<yarl>guix upgrde only for guix install

<ieure>folaht, You're mixing per-user profiles and Guix home profiles, which is causing your problem.

<yarl>you need guix home reconfigure after guix pull

<folaht>Okay, what is the difference between the two then?

<ieure>folaht, They are parallel systems, they each manage completely different profiles.

<ieure>folaht, So what's happening is, you have kitty in your guix home configuration and also in your per-user profile. The home config shadows the per-user profile, you didn't upgrade kitty in that, that's the version of kitty being used.

<ieure>folaht, Most users are best off using either a per-user profile or Guix Home, not combining both.

<folaht>Oh I see. I probably would want to use per-user profile, since I'm using this system on my home computer and there's always a new one coming.

<yarl>guix package/upgrade/remove (aliases to guix package) works on ~/.guix-profile

<yarl>(by default)

<folaht>And I have no idea where my list of packages even came from. I found it somehow.

<yarl>~/.guix-profile is a symlink to /var/guix/profiles/per-user/$USER btw

<yarl>guix home works on ~/.guix-home

<folaht>Is there something like a guix home-configuration template command?

<yarl>guix home import

<ieure>That.

<folaht>Hmmm... oh I see now what I've done.

<ieure>folaht, You can move packages from one to the other. `guix package --export-manifest' to get a manifest for your current PUP generation, `guix package -p ~/.guix-home/profile --export-manifest' for the Home package manifest.

<folaht>ieure, thanks. I'll look into that later. I was just curious how I got that package list in my home configuration file. Apparently I had done 'guix home import ~/src/guix-config' when I created my first home configuration scheme file and then expanded on that.

<folaht>Now I know what it does and that it needs to be removed.

<folaht>Is the home configuration meant for all users by the way? Like would it create the same ~/.bashrc file for all users including root?

<ieure>folaht, Not sure I'm understanding. Guix Home is a per-user thing, `guix home reconfigure' reconfigures the home of the user who invoked it.

<ieure>folaht, Multiple users can use Guix Home, you can share bits of configuration by putting them in a channel, but the output is always for the user doing the reconfigure.

<folaht>ieure, you're understanding it correctly. I was wondering if I needed a <user>-home-configuration.scm file for each user.

<tadhgmcdj>not if they all want to use the same configuration

<folaht>Oh right, so I can just use the same file for each user. I just have to invoke the same command on each one of them.

<ieure>folaht, The configuration itself doesn't matter, multiple users can share the same configuration, but the output of reconfiguring using it is always a per-user thing.

<tadhgmcdj>I don't think there is a system in place to automatically reconfigure the home config for several users at once, it is expected that users will just reconfigure their own home config as they need

<ieure>Like you could make /home/shared/home-configuration.scm, and `guix home reconfigure /home/shared/home-configuration.scm' as multiple users. As long as they can read the configuration, that will work. But the profile generation that creates is not shared.

<ieure>tadhgmcdj, There's a Guix System service which will deploy a Guix Home configuration, you can have multiple users' configs in there.

<tadhgmcdj>oh is taht supposed to fully deploy it? I thought that service just builds the home config and makes it ready to go for a new system

<folaht>My home configuration file has a local bash file though. how would that work in case there was some other user wanting to run 'guix home reconfigure'?

<folaht>Oh right, they'll probably would want to use their own .bashrc files, so I would better point those at ~/the/path/bashrc

<ieure>folaht, In my example, the local-file would have to be in /home/shared somewhere.

<tadhgmcdj>if you wanted seperate bashrc for each user and you do want that file managed by guix home then you inherently need different configs for each user, but how that difference is expressed doesn't have to be full seperate config files

<tachymelia>anyone else getting an http 504 on guix pull?

<attila_lendvai_>tachymelia, i get 4 out of 4 on an aarch64 box i'm just setting up, and 1 out of 1 worked on x86-64

<attila_lendvai_>i thought it's due to the aarch64 openwrt

<ieure>tadhgmcdj, Correct, you could have a conditional in the home.scm. But this is not a great option, it's better to do something like my profile approach, which encapsulates different parts of the config in transformation functions; then you select the transforms you want for a given outcome.

<ieure> https://codeberg.org/ieure/atomized-guix/src/branch/main/atomized/home/profiles.scm

<attila_lendvai_>tachymelia, is it on a machine that used to work? i suspect it's some missing https certs...

<ieure>attila_lendvai_, 504 cannot be a cert issue, if it was a cert issue, it wouldn't have been able to make the request the 504 is in response to.

<tadhgmcdj>is 504 the typical response when the host machine has their local time out of wack?

<tadhgmcdj>tachymelia: try running `date` and if that is obviously not correct it is likely the culprit

<identity>i doubt that, 504 is about relative time, not absolute time

<ieure>504 means the backend behind the reverse proxy didn't respond within the reverse proxy's timeout.

<identity>i doubt machine time is involved at all

<ieure>It means there's a Codeberg infra issue.

<ieure>Again, if there's a cert problem, it will not issue a request in the first place. 504 is a response to some request, thus, certs must be functional if you get one. Cert issues manifest as the software refusing to issue a request (ex a browser scare screen).

<attila_lendvai_>well, on my x86 3of3 worked. on this aarch64 openwrt i'm just setting up 6of6 got 504... this stinks more than a backend issue.

<tachymelia>attila_lendvai tadhgmcdj: yeah date's correct. it also pulled just fine from two other channels

<attila_lendvai>yay, the 7th pull went through on the openwrt, too

<tachymelia>I'm wondering if it's like. an issue w the git.guix.gnu.org redirect fucking up codeberg's ddos measures? or something like that?

<attila_lendvai>packet fragmentation and client speed may be a factor, but... that's just guessing

<attila_lendvai>tachymelia, that's a good hypothesis!

<tadhgmcdj>oh yeah, I remember having issues pulling on a new machine and then after accessing codeberg on a browser from the same IP it seemed to work, although trying multiple times may have been all it needed

<attila_lendvai>tadhgmcdj, heh, it was the same pattern here, so that will not make us smarter... :)

<attila_lendvai>i visited the website somewhere near in time when the pull went through. but then the x86 machine is on the same network, and it worked each time.

<tadhgmcdj>question, a while ago I posted a 1 line change pull request and 4 hours ago guix-cuirass-bot posted 24 replies with accompanying emails about build status.... https://codeberg.org/guix/guix/pulls/6303

<tadhgmcdj>does anyone know why that would happen or if that is expected?

<Rutherther>tadhgmcdj: sometimes the bot gets buggy, no it is not expected

<tadhgmcdj>ok but it is known, that is good enough for me

<tachymelia>ah yep visited codeberg and now it's pulling

<tachymelia>thanks y'all!

<sneek>wb janneke :D

IRC channel logs

2026-05-18.log