IRC channel logs

2024-08-16.log

back to list of logs

<oriansj>rgarcia`: you could just garbage collect that system generation
<ieure>issues.guix.gnu.org down -- known?
<ieure>Huh. When I run `guix style' on this Debian box, I get "guix style: error: mkstemp: Read-only file system." Anyone see this before?
<ieure>Oh, bizarre. It's trying to reformat files in the store, instead of $PWD.
<ieure>I guess this only works if I have a pre-inst-env set up.
<vagrantc>yeah, kind of the same problem as "guix edit"
<vagrantc>ACTION waves
<lh>ieure: ci.guix.gnu.org seems to be down too 😰
<Kabouik>I'm having issues trying to remove packages due to the ci.guix.gnu.org server being down. Is there a way to simply remove packages without that server?
<Kabouik>--no-substitutes, maybe? It just seems to take a while, I didn't expect it would need to build new stuff when removing packages.
<Kabouik>Actually guix remove fails even with --no-substitutes or --fallback on my end.
<Kolev>To be up to date, I have to do `guix pull` before `system reconfigure`, right?
<jack39>are there any specific donations that could help prevent downtime of ci.guix.gnu.org and issues.guix.gnu.org?
<jack39>newbie of me but anyone else needing substitutes right now i was able to add `--substitute-urls=https://bordeaux.guix.gnu.org` to guix-daemon and get things going for the moment.
<jaft_r>Kolev: correct
<Kolev>jaft_r, thanks.
<lh>my emails seem to have a large delay before showing up in archives, is there a moderator queue? is it just a certain number of allowed emails before you become allowlisted or is there some manual process I should undertake?
<Kolev>Question: Will my system work if this file system is not available? What if I unmoumt it? https://codeberg.org/csh/dotfiles/commit/dfef349417d318428229122fb0c87e548e0a5a67
<lh>it looks those are handled by non-boot-file-system-service in gnu/system.scm, which uses file-system-service-type in gnu/services/base.scm
<lh>Kolev: so it seems like the service for that filesystem would just fail if the device is missing
<lh>is geiser supposed to be able to complete all modules on your %load-path? for me I only get 146 modules from C-c C-m in a fresh repl, all of which seem to ship with guile
<Franciman>hi, I don't understand why this keeps happening and this morning the issue is worse than ever: https://bpa.st/S3DQ
<Franciman>as you can see, the name resolution fails saying that «fossilui.duckdns.org» resolves to 127.0.0.1
<Franciman>i keep invalidating cache, until once it gets it right and resolves to 144.91.112.31
<Franciman>but later, it keeps resolving it to localhost
<Franciman>and even now keeps failing. What's happening!
<Franciman>i have to say that this mostly happens with duckdns.org domains
<lilyp>for the record, a bunch of guix pages seem to be down for me, including issues.guix.gnu.org
<nckx>Everything on berlin.
<nckx>lh: There is, it's purged best-effort-daily, and it's per-list. Maybe that's why it seems you weren't immediately approved. Or your mails just got delayed unrelated to any moderation, that happens too.
<nckx>If your nick here is at all indicative of your name or e-mail address, no messages of yours are being held.
<nckx>As for berlin, I'm unable to connect to the other machine in the rack for management access, so either its configuration has changed (quite poss, it's been a while since I had to hard-reboot it) or the network is simply down, in which case it will simply come back up, whirh would be nice.
<nckx>I hope today is not a leap-holiday in Germany like it is here (lilyp?).
<lilyp>in parts of germany it might be – particularly bavaria, not sure about berlin
<lynn_sh>hello, im having an issue where i have setup swaylock but it does not accept the correct password. it also does not seem to have added an entry in /etc/pam.d/, any guidance?
<Rutherther>how did you try to set it up? the reason for not accepting the password is exactly that - you don't have a pam rule. It works only with a pam rule.
<lynn_sh>i just followed a basic wayland sway config i found on gitlab--im starting to think that may have been a bit naive; the sway and wayland part worked right out of the box but im guessing i need to make a service specifically for swaylock?
<lynn_sh> https://pastebin.com/BYSr4tT2
<lilyp>lynn_sh: The manual has an example using the ‘screen-locker-service-type’ – I guess this is probably exactly what you want 😉
<Rutherther>on other distros "xyz install swaylock" (xyz being apt, pacman etc.) usually also installs the pam rule. On guix putting stuff to packages can't create anything in /etc. So yeah, a service
<lynn_sh>ty both. im on day 0,5 here with guix so adapting is a bit new :)
<lynn_sh>i will add a service
<lilyp>note that `guix install' is way less intrusive than apt/pacman/etc. which typically require root permission exactly so that they can install pam rules and other stuff
<lynn_sh>lilyp i think perhaps i am misreading the documentation, or missing something else. on reconfigure it is telling me that screen-locker-service-type is an unbound variable (am i missing a package?) here is the system.scm https://pastebin.com/EhaBNWdM if you could take a quick peek.
<Rutherther>you are missing xorg in your use-service-modules
<lynn_sh>maybe this is naive, but i had figured xorg was unnecessary for wayland?
<rekado>reason for ci.* and issues.* downtime: the data center had a severe power outage, so there's not enough cooling capacity for all servers.
<Rutherther>yes, xorg is unnecessary. It's just that the definition of screen locker is inside of xorg.scm, I suppose for some historical reasons.
<lynn_sh>ah ok so i will include that, and maybe later add my own definition later when i am much smarter about this. thank you :)
<rekado>this happened about 9 hours ago. The people there are trying to fix it, get the temperature down, and then start the servers again.
<lynn_sh>ty for the update rekado i was wondering, i was installing this morning and ci was all down.
<rekado>all the build nodes have been shut off at 1am. At 5am the switches shut off, at 6am storage overheated and shut down.
<lynn_sh>Rutherther it all works. thank you for the help, i appreciate it.
<nckx>Thanks rekado.
<taeaad>There is already r-ade4, here: https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/statistics.scm#n2606. But I would also like to use adegraphics. Do I have to figure out all the dependencies, repo etc. or could there be a shortcut to have it work similar to r-ade4, for example I see r-build-system and url-fetch, nothing more complicated.
<taeaad>For example, here: https://cran.r-project.org/web/packages/adegraphics/index.html it says that adegraphics is at the URL http://pbil.univ-lyon1.fr/ADE-4/, but this is the same URL used for r-ade4.
<Kabouik>I think the Guix package for ade-4 pulls from (uri (cran-uri "ade4" version)), https//pbil.univ-lyon1.fr/ADE-4 is the home-page url.
<Kabouik>Most likely you would need to create a separate package for adegraphics if r-ade4 does not include the corresponding R packages (I havve not been using ade4 graphics since I moved to Guix, not sure if adegraphics is bundled with ade4).
<taeaad>Kabouik: It doesn't look like adegraphics is installed with it. I can do library("ade4") but not library("adegraphics").
<Kabouik>Then I think you need a separate Guix package. You can start with `guix import cran adegraphics` to avoid the white page anxiety. There is also a Guix CRAN channel at https://github.com/guix-science/guix-cran.git. I use it, but I also get version conflicts with my R packḱages quite often because I think I'm mixing packages from the main Guix channel and that one, I'm not sure how to pull R packages only from that one.
<taeaad>How do I reference the R executable or should I leave it to url-fetch?
<taeaad>For example it is confusing to me that https//pbil.univ-lyon1.fr/ADE-4 is the package URL if it is a sort of human directed homepage with links, in fact, actually it redirects in your browser to https://adeverse.github.io/ade4/.
<taeaad>I seem to be able to do guix build -f r-adegraphics.scm. How do I then install it?
<taeaad>I mean, install it to my profile?
<rekado>the home page is taken from the DESCRIPTION file
<Rutherther>taeaad you can use guix package --install-from-file
<lilyp>what's our state of meson one core-updates?
<taeaad>Kabouik, Rutherther: Looks like it worked, yay.
<taeaad>rekado: Does that come from CRAN?
<taeaad>For example I saw it said: substitute: updating substitutes from 'https://bordeaux.guix.gnu.org'
<Kabouik>Good news taeaad!
<Kabouik>Why does `guix remove xxx` need to build stuff? This makes the whole thing very time-consuming.
<civodul>not sure what happened but thanks to everyone who took care of bringing ci.guix.gnu.org back up!
<civodul>ACTION restarts x86_64 workers
<civodul>avahi-daemon occasionally fails to restart due to stale PID files
<Kabouik>I don't understand why `guix remove distrobox podman` is rebuilding the world on my machine, including thinks like gtk, vulkan or python-*
<Kabouik>things*
<janneke>ACTION finds why hurd-team's `devel-hurd.tmpl' didn't build on core-updates+patches -- a new imagemagick dependency
<civodul>hmm, looks like kreuzberg didn’t come back
<janneke>(and imagemagick's dependencies don't cross-build for the hurd atm)
<civodul>janneke: d’oh
<civodul>my childhurd ran out of space, i cannot build /gnu/store/hwd1jc5vlsbg8z9nda5q8148dkcfwbi2-gcc-11.4.0.drv
<lilyp>Kabouik, did you recently gc?
<janneke>ACTION usually uses 25G childhurds
<taeaad>I thought no-one used hurd.
<janneke>civodul: yeah; anyway just reconfigure'd on core-updates+patches; childhurd works for me \o/
<janneke>(including the development packages from https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/system/examples/devel-hurd.tmpl?h=hurd-team, minus imagemagick)
<janneke>taeaad: hurd is in development, very few people really use it yet
<Kabouik>lilyp: yes I did, I needed space, ut thought removing packages would not need to rebuild
<nckx>'Removing a package' = 'building a new profile'.
<nckx>Profiles require a surprising heap of software to build.
<rekado>civodul: info on what happened: https://logs.guix.gnu.org/guix/2024-08-16.log#103139
<nckx>Kabouik: --gc-keep-{outputs,derivations}=yes probably mitigate this somewhat, at the expense of just keeping more stuff live.
<nckx>Although ci.guix.gnu.org 'cannot be rated' and is hence inaccessible, I hope the censors get to it soon.
<Kabouik>I see, thanks
<Elouin>Will any guix folk be at froscon this weekend? Even maybe a talk about guix?
<Kabouik>That will come in handy nckx, thanks. I'm on a mobile device, so both space and build time are limiting, interesting to see that alternate gc argument.
<nckx>I hope it helps (I use it everywhere so I don't remember much about life without it). It's a guix-daemon argument.
<Kabouik>I'm actually not sure how to change guix daemon arguments on this foreign distro machine. The machine runs Droidian (Debian on Android kernel), so I guess it is done in /usr/lib/systemd/system/guix-daemon.service?
<rekado>nckx: what does "cannot be rated" mean?
<Kabouik>I guess this should also considerably accelerate all my guix install/guix remove commands. I'll try it nckx, just hope it won't eat too much space because sometimes guix gc was saving me when out of storage on the phone.
<nckx>rekado: It was some WAF thing, I can't reproduce it.
<nckx>Kabouik: What OS does this 'phone run? Sounds fun.
<Kabouik> https://droidian.org/ nckx, and it has a 66-keqey sliding keyboard too, so it's a nice pocket PC with telephony
<Kabouik>66-key* even.
<civodul>janneke: yay, thanks for testing!
<civodul>(sorry, went for a walk 😎)
<Guest98>Has anyone experienced certain actions causing guix to crash to login screen?  Opening some links will cause the crash for me, and trying to open a video file has also caused it for me as well.  Any ideas?
<Guest98>this hasn't happened to me before on other distros, so I think it's guix related.
<AwesomeAdam54321>Guest98: Does xdg-open cause the crash? Usually `xdg-open video-file.webm` would open it in the default video player
<Guest98>AwesomeAdam54321 i'm not sure, but i tried opening it before from within vlc, which is not my default player, and it still caused the crash.
<Guest98>AwesomeAdam54321 the package for xdg-open wasn't even installed on my system, so I'll see if that changes anything
<Guest98>AwesomeAdam54321 xdg-open fails to open the files, but doesn't crash, so it doesn't seem to be related.
<guestinator>how can I unquote a variable in a gexp? I have a procedure defined in my module that I want to replace a phase with
<guestinator>ungexp (#$) does not seem to work, I get invalid G-expression input. tried various combinations of gexp and quasiquote but couldn’t get it working
<guestinator>e.g. `#~(modify-phases %standard-phases (replace 'install ,my-install-procedure)) gives: Unknown # object: "#<"
<lilyp>`#~ is overkill, you do (list #:phases #~(...))
<guestinator>lilyp: then how do I unquote my procedure variable? that gives me ‘unquote: expression not valid outside of quasiquote in form (unquote my-install-procedure)’
<guestinator>or do I have to export my procedure in a separate module and use that module within the gexp in order to accomplish this
<guestinator>right now I just have a (define my-install-procedure …) in the same file as the packages I’m using it in
<guestinator>well, trying to use it in, hah
<rynn>Is there a way to see which packages you've installed for a given channel?
<lilyp>guestinator: what about #$my-install-procedure?
<guestinator>that gives me ‘guix build: error: #<procedure my-install-procedure args>: invalid G-expression input’
<rynn>I did find the --list-installed option for guix package, but that only gives the name and store location, doesn't seem to be any additional flags that can be added that give the channel
<taeaad>janneke: Do you use hurd?
<guestinator>lilyp: I was able to fix it by changing my procedure variable to just quote the entire thing (i.e variable just holds a sexp) and then #$ worked fine. doesn’t seem like the best solution but at least I can reuse the definition now
<civodul>janneke: i’ll upload the glibc-bootstrap tarball and push the patches
<civodul>i’m glad you were able to test, ty!
<janneke>civodul: (y)
<janneke>taeaad: no, i only do some packaging for guix
<eikcaz>Suppose I enable substitute sharing with guix publish. Is there an easy way to have my substitute server build missing items when they are not found? I'd really like to avoid having my puny ARM device try to build anything.
<eikcaz>I've accomplished this manually by running guix commands on the server, and doing a guix copy from the arm device, but I'm hoping there is a more "automatic" way
<eikcaz>Ah, the answer was in build daemon settings
<janneke>eikcaz: you can offload builds to your server
<janneke> https://guix.gnu.org/manual/devel/en/html_node/Daemon-Offload-Setup.html
<lispmacs[work]>hi, a shepherd service is not starting but not giving a reason why. can I pass an option or something to get the reason?
<civodul>lispmacs[work]: are there any more details in /var/log/messages or in the service’s own log file?
<lispmacs[work]>no, not initially, anyway.
<lispmacs[work]>after I rebooted the system, I started to see helpful messages in the services log file
<User-39>When I add a service like prosody-service-type, there should be a command prosodyctl, but command isn't found. I've tried with sudo too with the command and same thing. Am I supposed to be adding the package to system packages too for it to be available?
<lispmacs[work]>so, crisis averted. i am a little curious still though if herd has some built in logging or verbose mode that tells you error code or something
<ieure>User-39, Yes, adding services doesn't add the package to your profile.
<User-39>Thanks. I thought adding the service would make the commands be available too for it.