IRC channel logs

<postroutine>There is something I'm not certain to understand about the `mumble-server-service-type`.

<postroutine>The data type `mumble-server-configuration` use the package `mumble` by default.

<postroutine>But, as this package contain the server and also the client: If I add a `mumble-server-service-type` to my server, will I end with a full graphical stack ?

<postroutine>As the mumble client is a GUI application .

<Guest3725>How do I get Guix to look for keys only in `~/.ssh/`? Right now it's looking in my `config.scm` directory, which I don't want.

<adanska>Hi guix!

<yarl>Hi adanska

<adanska>whats up yarl :)

<yarl>Waking up ;)

<adanska>relatable... except for the fact that its 2pm here and i allegedly woke up 4 hours ago

<yarl>That's why I'm takink a big bowl of black coffee.

<adanska>ive got my green tea and prescription adhd meds so taking similar precautions

<dajole>Is there a way to check the output of `local-file` in the Guile repl? It seems to return some sort of record (still learning scheme) instead of a string.

<dajole>Hm...`(lower-object (local-file...` gets me a procedure...

<dajole>Ah...I think using `guix repl` and then `,build (local-file...` gives me the result...maybe?..

<test202020>ieure: last difference in home directory between users is guix home configurationx but i not xhange that before update to gnome 48

<test202020>very strage how that broken gdm login

<test202020>because after login to broken user home gdm cannot recive password for two user too. need to kill gdm for login to second us€r

<test202020>right, somethink broken in guix home reconfigure...

<futurile>Morning all

<test202020>hi)

<adanska>is there really no way to have cross-builds with the pyproject build system?

<adanska>it essentially kills the prospect of cross-compiling any sort of more-than-trivial system image

<adanska>it sucks because having to emulate aarch64 to compile something like a custom kernel is extremely slow

<adanska>and theres no way to mix cross-build artefacts and native artefacts

<Rutherther>adanska: there is a way to mix cross build artefacts and native artefacts. Right, no one implemented cross builds for pyproject build system yet

<adanska>Rutherther, what is the way? I'm very curious because it would make my life so much easier.

<Rutherther>adanska: https://paste.debian.net/hidden/a1f55dc5 this will always compile package natively for aarch64-linux

<Rutherther>if you instead set the %current-target-system to something and left out the %current-system completely, it would always cross compile

<adanska>Rutherther, oh my goodness I have to try this!!!!!! this is a massive help thank you so much :DDDD

<adanska>Rutherther, I've been going around and fixing a bunch of cross-compiling problems trying to find a happy path for my system build, but this way I can just cross-compile the builds I don't have subs for!

<adanska>Rutherther, I'm having trouble using `with-parameters` inside of a operating-system declaration. I'm trying to wrap my kernel package in it.

<adanska> https://paste.debian.net/hidden/04f0cef7 here is the error, it doesnt really make things any clearer

<Rutherther>it will probably not work for the kernel package directly due to how it's used in other services

<Rutherther>are you using inferiors?

<kestrelwx>Hello!

<adanska>Rutherther, no i'm not. I have also tried making a seperate package symbol with `with-parameters` wrapped around it and it doesnt seem to help.

<Rutherther>adanska: it will just not work. The kernel package is handled very specially by other services. It needs to be a package, not an arbitrary gexp input. So your only way would be making a package wrapper that will copy the outputs of the cross compiled kernel

<adanska>Rutherther, I see... seems clunky, but honestly it would be worth it. I'll look into this. Thanks so much for your help :)

<simendsjo>cutlass is acting strange in a PR. Who should we tag to help look at it? https://codeberg.org/guix/guix/pulls/6805#issuecomment-11630278

<simendsjo>s/cutlass/cuirass

<csantosb>You can try: https://codeberg.org/guix/cuirass/issues

<yarl>hi

<apteryx>this feels retro: playing a DVD with VLC on Guix System. The seeking noise is a big factor of the nostalgia ;-).

<civodul>apteryx: heh :-) i don’t even have a working drive left

<apteryx>I just bought a slim external one which I plug via USB (I think it can play every optical disc CD, DVD, Bluray). It supposedly can write bluray and dvds (not sure about CD-RW). I hear optical discs have good bitrot resistance, maybe I'll try burning a bluray with some photos.

<fhang>apteryx: That's the reason why I still use optical discs for archival backups.

<n|Phreak>curious , pretty sure I'm not using networkmanager even though its enabled in hurd , wpa_supplicant works just fine even after reboot , but dhclient has to be run after reboot and so does unlocking wlan after some reboots

<coopi> https://bpa.st/G5AJM

<coopi>I'm trying to make a Guile script for my Guix dotfiles

<coopi>I'm getting a backtrace but I don't understand what I'm doing wrong

<coopi>What's weird is that the script was working before; I left for like five minutes and then it stopped...

<coopi>I'm not sure if I did any changes before leaving; though stepping through Emacs' undo history, I couldn't find anything incriminating.

<n|Phreak>Ok , so I was right , once I installed nm-tray , I'm now using networkManager now

<rovanion>I'm reing in the manual [1] that if I get failures when running `make check` I should email bug-guix@gnu.org - is that still the case?

<rovanion>[1]: https://guix.gnu.org/manual/devel/en/html_node/Running-the-Test-Suite.html

<Rutherther>rovanion: no, report it as an issue on codeberg (ideally after looking there is not an issue for it already)

<Rutherther>the manual should be adapted

<n|Phreak>So I'm looking for information for herd like systemctl for systemd , like how to find errors or debugging a service ?

<rovanion>Oh, `make check` is known to be broken... https://codeberg.org/guix/guix/issues/4225

<n|Phreak>do you have to dump a running service in order to see whats going on ?

<mwette>sneek, later tell coopi: I think you need to export `main'

<csantosb>andreas-e: Hi Andreas ! I'm looking at qa.guix.gnu.org/branch/hpc-team, it is great to have the json view

<csantosb>I'm trying to figure out the meaning of "unknown", paste.sr.ht/~csantosb/f84c487981a2f6dd2e63783355ad8e7e6f41cd7b

<Rutherther>unknown means that it hasn't been built yet, so the state is not known

<csantosb>Last one in the json file, "emacs-casual-calc", both target and base show "status": "succeeded"

<Rutherther>csantosb: so what was your query then?

<csantosb>Forget about, sorry; I probably mixed json files somehow

<Rutherther> https://data.qa.guix.gnu.org/compare/package-derivations you can read the meanings here

<csantosb>I see, thanks; the json view is way more useful than the web ui

<Rutherther>you can use the web view to filter and then just stick in .json

<csantosb>Yes; now, I just need to convince librewolf to open json with an external program

<csantosb>"potentially_blocked" and "build_for_equivalent_derivation" are still a bit obscure to me

<yelninei>ACTION rushes to text guix pull on x86_64-gnu

<rgarcia`>hi! I'm trying to create a package. I've added it to a local guix clone. But when I ./pre-inst-env guix lint it cannot resolve hostnames. Help!

<andreas-e>csantosb: I do not know what has happened, rebasing seems to have set back the hpc-team branch quite a bit. Now it is at 2000 packages still to be built for x86_64. But the build farm is running. Also, lots of new commits on master, which take precedence.

<rgarcia`>I guess pre-inst-env overwrites some variable, maybe I should install something into that shell?

<csantosb>rgarcia: You did a `guix shell -D guix` before ?

<csantosb>andreas-e: Yeah, I was a bit surprised too

<andreas-e>I think we did not rebase during one week, and maybe this is just the normal churn rate of master...

<andreas-e>And since hpc-team is close to a world rebuild, we need to rebuild more or less everything that changed during this week on master. Time to merge :)

<csantosb>Well, world rebuild means close to 30k, right ? We're at 1/3 of that.

<jlicht>is there any particular reason I'm missing on keeping our sqlite "so old"? guix refresh lists just a couple of packages depending on it, but perhaps I'm not seeing the forest for the trees :)

<yelninei>jlicht: Building the following 6805 packages would ensure 23617 dependent packages are rebuilt for sqlite@3.39

<csantosb>We have also sqlite-next

<andreas-e>jlicht: This is the trap of "guix refresh sqlite", which takes the last version of the package with name field "sqlite", which is in fact the variable sqlite-next.

<andreas-e>We should do a real world rebuild with an update, and not just the hpc-team pseudo world rebuild :)

<dajole>I'm trying to get `gpg --card-status` working. It reports that the service is not running. I added pcscd to my operating system config and it's also running. How do I figure out how to debug this further?

<dajole>Hm...`sudo herd status pcscd` shows `../pcsc-lite-2.4.1/src/auth.c:166:IsClientAuthorized() Process 768 (user: 1000) is NOT authorized for action: access_pcsc`

<dajole>Seems like some sort of permission issue, but I'm not sure where to go from here.

<Rutherther>dajole: what did you add to your operating system config?

<dajole>Just `(service pcscd-service-type)`

<Rutherther>did you reboot afterwards?

<dajole>yexs

<dajole> https://guix.gnu.org/cookbook/en/guix-cookbook.html#Using-security-keys mentions udev rules, but it's unclear to me whether this should be needed for `gpg --card-status`.

<Rutherther>dajole: are you using either of elogind (it's in %desktop-services) or seatd?

<dajole>No, this is on a fairly bare-bones home server.

<Rutherther>okay, I think that's the issue. The policy uses <allow_active>yes</allow_active> which presumably uses elogind to check for active sessions

<Rutherther>(or rather it uses dbus to talk to a protocol that elogind understands and replies to)

<dajole>Interesting, at least that gives me something to investigate. Thank you! Out of curiosity, and for the future, what made you think about that specifically?

<Rutherther>it's a polkit policy declining you, so to look for answers I looked inside of it - "$(guix build pcsc-lite)/share/polkit-1/actions/org.debian.pcsc-lite.policy"

<Rutherther>and I see inside of it allow_active yes. Which wouldn't work without elogind, so question about elogind naturally emerged

<Rutherther>you can change the policy to allow any, that would fix it

<Rutherther>or you can use elogind

<dajole>I see, I'll investigate further, awesome, thank you!

<bjc>are the container tests hanging forever for others?

<bjc>looks like it wedges in the "call-with-container, pid namespace" test for some reason

<bjc>^C won't kill it, either. ‘strace’ shows it stuck waiting on a futex

<yelninei>janneke: I ran guix pull --no-grafts --url=$PWD successfully now on x86_64-gnu (got scared of a gcc derivation so --no-grafts), "only" took ~2h with all dependencies already built

<bjc>wow

<tarcv>Hi. Do mails from proton.mail blocked on guix-devel list? Wanted to share that I got Gradle 4.5.1 bootstrapped, but see no message in the archive.

<jlicht>ah the sqlite-next thing, thanks

<rovanion>I've managed to break something here. By changing the version of musescore I... got what I think is the current master branch downloaded and compiled and nothing cared about the mismatching sha256-sum. https://paste.debian.net/hidden/6adcc025

IRC channel logs

2026-03-14.log