IRC channel logs
2026-05-07.log
back to list of logs
<nutcase>GUIX_TEXMF is /home/flake/.guix-home/profile/share/texmf-dist:/home/flake/.guix-home/profile/share/texmf-dist in my home profile (yes, it doubles the texmf-dist directory. <ekaitz>nutcase: basically because the container doesn't give them access to the whole directory... they still try! <nutcase>ekaitz: my container has --share=/home/flake <mwette>bdunahu: I got prusa-slicer to build after patching cereal, embree, embree-3 and libigl! (libigl had me confused for a bit) I submitted update for cereal. <nutcase>ekaitz: WDYT, is it worth to open another issue to ping the tex team? <nutcase>or can we use your #6317 to ping them? <ekaitz>you can reopen and maybe ping andreas yeah <mwette>bdunahu: I have not tried yet. I'm away from that computer (aarch64 desktop) right now. <nutcase>I pinged enge, but it seems I'm not allowed to reopen someone else's issue. <ekaitz>well, it's not an issue of the package really <nutcase>ekaitz: ok. thank you very much for your immediate support. the strace helped to narrow down things. I'll see, whether enge has an idea <ekaitz>nutcase: no problem! i would just try to read the latex font package you use and see how it tries to match the font, what font you have actually installed in your system (use fc-cache or fc-list or whatever that is for that) <ekaitz>you might realize it's the latex package dependency that is broken <ekaitz>and that's why it might not be found <mwette>bdunahu: prusa-slicer window comes up, but program crashes on "Failed to load credentials from the system password store." <bdunahu>maybe it can't access gnome keyring or something? just guessing <mwette>internet says that is non-fatal. Must be something else. I'll need to buidl for debugging, likely. <mwette>ldd shows bound to 225 lib's from /gnu/store <apteryx>we now have emacs-forgejo in Guix, if someone wants to try. Perhaps a could complement to emacs-fj; I haven't tried it yet. <sham1>It doesn't really look to be like Magit, it's more like Magit's Forge <sham1>So doing things like browsing issues, doing PRs, and that sort of stuff <gabber>just upgraded my laptop to a system where i don't have to enter the decryption password twice. now my wifi doesn't connect. nmtui reports: "Could not activate connection: Activation failed: Serets were required, but not provided" <gabber>the wifi password did not change and is still set when i open the edit menu in nmtui <csantosb>Emacs people: with guix-home, ~/.authinfo.gpg becomes a symlink to an store item <csantosb>This seems problematic, (auth-source-pick-first-password :host "codeberg.org") returns nil; any idea ? <apteryx>did you restart emacs/drop the authinfo cache? <csantosb>Even tried with "(auth-sources (list (file-truename "~/.authinfo.gpg")))" <apteryx>I use a symlink since a long time and it's fine: /home/maxim/.authinfo.gpg -> stow/emacs/.authinfo.gpg <apteryx>did you review the .authinfo.gpg content? <csantosb>This is because you're using stow, not guix-home <apteryx>I know; I'm just saying a symlink is not the issue <csantosb>You're pointing to a .gpg file; I'm pointing to "...svcyc0wi-home-dotfiles--authinfo-gpg" <csantosb>Which explains why "(file-truename "~/.authinfo.gpg")" doesn't solve it <apteryx>you could try setting auth-source-debug to someth6g like trivia to get debug messages <apteryx>I also do not have special auth-source config in my .emacs <civodul>csantosb: i use Home but ~/.authinfo.gpg is not a symlink to a store item; perhaps you added it to the list of home-dotfiles-service-type? <gabber>csantosb: i also use home but i don't let it manage ~/.authinfo.gpg <gabber>but my setup is somewhat cursed anyways, so i might not provide the most valuable input <csantosb>I need to manually remove the symlink to make it work <csantosb>apteryx: this 'trivia setup helps to understand what's going on here, thanks ! <csantosb>By the way, new emacs-forgejo around, fixes a couple of bugs with replies <sham1>Is there some kind of comparison between emacs-forgejo and emacs-fj <sham1>Like how do they actually differ from each other <csantosb>I'm pretty sure I read that Thanos is for merging the two packages <theesm_>good morning guix o/ does anyone else also have the issue that codebergs notifications are too broad? (ideally i'd only get mentions and team-kernel/team-perl mentions, but currently i seem to get all issues/PRs which leads to me missing pings) <csantosb>theesm_: check the 'watch' button in the codeberg interface <theesm_>csantosb: thx! it was set on watch for the repo (which explains the amount of notifications) <gabber>upgrading the system on another laptop (pinebook pro) i get: "build of /gnu/store/pars71272khvg4kqvb040bh5h3qkfglq-system.drv failed" with the content of the log file: "ERROR: In procedure symlink: In procedure symlink: File exists" what am i doing wrong? <gabber>i guess there's a symlink for the system generation to be created already existing..? is it because i pass `--allow-downgrades`? <trev>anyone else use zsh and notice a "/etc/profile:68: no matches found: /etc/profile.d/*.sh" error? <gabber>bordeaux is very slowāthat is normal, right? <trev>csantosb: never seen that error on login shell? <csantosb>I don't remember now; I'll check that tomorrow when running guix system. <yelninei>civodul: Looking into updating binutils, glibc, zlib and libgc <gabber>how can i turn off specific test in a rust crate? they seem to depend on image data that does not seem to be delivered with the sources <yelninei>Updated binutils failed against updated glibc ... Great <atw>Hello! If you use emacs-magit, then while browsing Info manuals in emacs, you may have seen "apply: Autoloading file /gnu/store/ā¦-emacs-magit-4.5.0/share/emacs/site-lisp/magit-4.5.0/magit-base.elc failed to define function Info-follow-nearest-node--magit-gitman" when you try to click links. I have a workaround: (advice-remove #'Info-follow-nearest-node 'Info-follow-nearest-node--magit-gitman). <civodul>woow, i havenāt experienced it, but it looks like you digged quite a bit :-) <atw>But, I don't know exactly why this happens. Doing C-h f on Info-follow-nearest-node shows that it has 2 pieces of advice, one which seems to work and one that doesn't (which we remove with advice-remove). I'm mentioning it in #guix, though I do not know if it is a Guix problem... kinda guessing here but maybe it's something to do with Guix's byte-compilation? <apteryx>civodul: would you have an idea as to why, if /gnu/store is bind mounted from elsewhere, e.g. /mnt/tmp/gnu/store as is the case in the hetzner rescue for our 'guix deploy', guix-daemon stops working? I guess layering bind mounts is a no-no ? <civodul>atw: maybe open an issue and tag @guix/emacs for their advice? (pun not intended) <civodul>apteryx: stops working how? bind mounts are āinvisibleā so i would expect it to be fine <apteryx>and "breaking" gnu-store.mount by setting Where= to an invalid value happens to fix it, as happens with the "fix" in 3e1befe1d8e <atw>civodul: I could, though I'm not sure this happens on machines apart from mine... I was hoping to get a "hey me too!" before I say that this is definitely an issue <civodul>atw: i donāt have problems clicking links in Info manuals <yelninei>binutils failed again ... (the upstream fix is 5 days old) <civodul>apteryx: i donāt quite understand 3e1befe1d8e3e1fae38abd8b5f89db2cd764372b <apteryx>civodul: guix-daemon just fails to exec; the output doesn't really give any clue <civodul>apteryx: guix-daemon cannot be executed, or it fails to execute other programs? <apteryx>civodul: its cryptic, but the result is that it seds the Where line in gnu-store.mount to be Where=//mnt/tmp/guix/store instead of Where=/gnu/store, before running 'systemctl daemon-reload' in our install script. <apteryx>I thought this made sense at first, but it doesn't; Where is the destination, and it's used as the source here. Also systemd insists on the file name used in its 'Where' matching the file name in its name, which is what breaks it. <apteryx>it does fix 'guix deploy' on hetzner though, so it buys us time to figure out a clean fix ;-) <apteryx>the hetzner script (hetzner-machine-rescue-install-os) does mount --bind /mnt/tmp/gnu/store /gnu/store before running guix-install.sh <yelninei>i have the feeling a lot will be broken by -Wdiscarded-qualifiers :( <apteryx>civodul: to answer your earlier question which I had missed, guix-daemon cannot be executed at all <apteryx>systemd says: (code=exited, status=203/EXEC) <apteryx>and then confusingly I think /var/guix/profiles/per-user/root/current-guix/bin/guix-daemon does not exist; /var/guix/profiles/per-user/root/current-guix is an empty directory <apteryx>is this populated by guix the first time it runs or something? <apteryx>can't be, since the ExecStart wants it there, so it should exist <civodul>apteryx: an AppArmor policy issue maybe? is that Ubuntu? <PotentialUser-64>has anyone compiled a rust project to wasm with the cargo-build-system? (Is it even supported?) <attila_lendvai>elogind has system-sleep-hook-files and system-shutdown-hook-files, but i want to add a call to efibootmgr -n 0003 when hibernate happens, so that the bios boots linux after a hibernate, not the default boot entry. is this just a TODO? <attila_lendvai>if i have a plain-file in my config that contains full store path references in comments, will that bring in the dependency in a system recofigure? <attila_lendvai>my ultimate goal is to add a hook to /etc/elogind/system-sleep/ through system-sleep-hook-files, but i'm groaning in escaping hell... <attila_lendvai>the shell-script-in-a-plain-file variant is readable/bearable, but i need a gexp in a program-file to reference packages with #$(file-append efibootmgr "/sbin/efibootmgr") <ieure>attila_lendvai, Not sure I understand what you mean about bring in the dependency in a system reconfigure. If your plain-file references a store item path, the referenced store item(s) are considered part of the closure for your plain-file. So `guix gc' won't collect those. <ieure>A comment isn't part of the input graph, but I'm not sure whether Guix will build derivations mentioned in comments in a store item. <attila_lendvai>ieure, that's what i meant, thanks! so, the gc scan goes through the bits mindlessly looking for anything that looks like a store path, right? <attila_lendvai>maybe i can build the body of the plain-file with some string-append. that won't be pretty either, but at least i won't go through a guile layer <ieure>attila_lendvai, mixed-text-file is likely what you want. <ieure>attila_lendvai, Awesome, glad to hear it! You're very welcome. <attila_lendvai>yay! no more struggle with the bios booting the default EFI entry without any sane way to stop it.