IRC channel logs
2024-04-25.log
back to list of logs
<ulfvonbelow>Kolev: in those situations, you set the 'method' field of the source origin to url-fetch/tarbomb instead of just url-fetch <zilti>Something else I've been wondering is, if one package is a plugin to another, what do I have to do so it is visible to it? <ulfvonbelow>look for an environment variable that can specify the plugin search path, and add that to the native-search-paths of the main package <ulfvonbelow>if no such environment variable exists (for example, the main package is hardwired to search a single directory somewhere under its install prefix), you may have to write a patch to add it yourself <zilti>But I don't know ahead-of-time what path the plugin will be installed in? <ulfvonbelow>right, the search path specification doesn't use the actual install prefix <ulfvonbelow>the short version is that you specify something like lib/somepackage/plugins, and then whenever an environment is constructed for a profile containing that subdirectory, a corresponding environment variable definition will be set <PotentialUser-79>Do you know how can I install trash can on xfce? I already installed gvfs <zirpu>the fish package is available. guix search fish <PotentialUser-79>I'm asking because I want to use tmux and have ' set-option -g default-shell "home/<user>/.guix-profile/bin/fish" ' in .tmux.conf and it doesn't work for me. What is interesting, is that for alacritty the config ' program = "home/<user>/.guix-profile/bin/fish" ' works just fine. <zirpu>what's the value of the SHELL environment variable? probably the full path to bash. <zirpu>you could restart tmux with SHELL=.../fish and see if that works. <zirpu>also, you should probably have it as /home/<user>/.guix-profile/bin/fish needs to have the leading / there i believe. <PotentialUser-79>Yes, in config I have leading /, but I didn't rewrite it correctly in libera.chat <PotentialUser-79>zirpu is it possible to have shell point to newest fish? Because if I would upgrade fish, the link will be broken <zirpu>point to the symlink in $HOME/.guix-profile/bin/fish that should probably be the latest. I'm fairly new to guix myself, so not yet 100% certain about whether the old installs go away immediately or must be garbage-collected. <PotentialUser-79>zirpu it still doesn't work. When I open alacritty and "echo $SHELL" I get /home/<user>/.guix-profile/bin/fish and when I open tmux in same session, I login into bash, even if when I run "echo $SHELL" I still get path to fish. <PotentialUser-79>I really don't understand why tmux doesn't like using different shell. Maybe this is because of guix different file system <zirpu>tmux should be using the $SHELL env-var value. or your .tmux.conf value should override it. this is tmux 3.3a? <zirpu>if you open a new tmux window, what is the value of $SHELL there, bash or fish? if it's fish, then tmux is ignoring both $SHELL and the .tmux.conf setting. <zirpu>so it's using the `default-shell` value or the env-var? <PotentialUser-79>Now shell is bash for all compatibility reasons and tmux uses fish with just set-option -g default-shell "/home/<user>/.guix-profile/bin/fish" <apteryx>there are some exciting contributions on the patches tracker! <apteryx>guix 1.5 -- speaking for myself i'd be working on this after the core-updates branch is merged, but there's no hard schedule. probably months <apteryx>not sure if it's exciting to others, but the inability to mount NFS at boot has been a long pain point for myself :-) <PotentialUser-79>Cool, good luck with work! I'm going and I don't know how to save irc, so bye <PotentialUser-60>Hello. I wonder why python-jaxlib has no substitutes for quite some time. Is there a reason for that? <futurile>PotentialUser-60: uh is it called that - I don't have anything for 'guix search python-jaxlib' <futurile>cbaines: looks like qa is having a bad day <futurile>PotentialUser-60: no worries, I guess ask on their bug tracker <PotentialUser-60>guix data service is also not well responding, but that might be my employees bad internet connection <futurile>PotentialUser-60: I think you're right - data service, qa etc look like they're having some issues <PotentialUser-79>When I run commands in qemu vm that takes long time, like guix pull, after some time qemu goes black and stops commands if I don't move mouse over qemu window. Do you know what can I do to fix that so I can run command and go? <decfed>anybody know of a way to run org source blocks inside a guix shell environment? <jakef>decfed: i know how to run it in a guix profile but not a shell. that would be cool <PotentialUser-79>and type sudo at the beginning? The && won't work, because I want to type password before guix pull starts <morganw>The easiest way, if you know that guix pull will not take long enough to require entering the password again, would be: <morganw>sudo true && guix pull && sudo guix system reconfigure /etc/config.scm <morganw>Whether that is good, I'm not sure. I guess it is safer then disabling the requirement for a password when using sudo, but would not be reliable. <morganw>(true does nothing and returns as successful) <morganw>I imagine you could also configure sudo to not prompt for password when running the guix command, although maybe there would be security implications in doing that. <PotentialUser-79>First guix pull took me like 1h, so I don't think sudo will be remembered for that long <zirpu>I would just do sudo -s and run it from within the root shell. then exit after done. <morganw>I don't think the pull is meant to be done as root. <PotentialUser-79>I now get error "guix substitute: error: TLS error in procedure 'write_to_session_record_port': Error in the push function" with "...-guix-1.4.0/bin/guix substitute died unexpectedly". Do you know what should I do? <jakef>repeat the command. it should work or get further this time. there's an issue floating around about that <podiki>cbaines: the us mirror (I'm using it again) gives lots of 404s for nars (and this is on a current guix, not time machine) <minikN>Hello, has anyone experience with mounting smb/cifs share in guix? As in write a file-system expression for it. I'd appreciate any help. <tschilp>Hi guix! I'm having some trouble with tramp connecting to through orgmode. I'm trying like this: http://paste.debian.net/1315182. But I receive '/gnu/store/.../bin/sh: no such file or directory'. I've already set '(add-to-list 'tramp-remote-path 'tramp-own-remote-path)' in my init file (as I do when connecting from debian to guix, but guix to raspberry pi os seems to cause trouble at the moment). Any ideas? <tschilp>this most certainly has worked before, so I guess there've been changes (although the tramp manual seems to have stayed the same). <ieure>tschilp, Issue is that $SHELL from your local machine (which points into /gnu/store) is getting sent to the remote, where that file doesn't exist. <tschilp>ieure: do you have an idea how I would prevent this from being done? <ieure>tschilp, I think you need to set shell-file-name. I know you can configure SSH to send/not send environment variables, but I'm not sure if you can tell it to only send certain ones, or not send $SHELL. <aitzkora>I want to load a local module in the REPL, is there kind of -L . flag in the the guix repl command <decfed>I cannot understand search paths. I would like to add library code for chez scheme to its search path, which in the package chez-scheme is defined as CHEZSCHEMELIBDIRS. However when I run guix shell chez-scheme --search-path this variable is not listed. <tschilp>ieure: cool, thanks for the hint -- in this case here, I would not need any! <decfed>Hmm, okay the path is set when I ask for another library to be included in the shell as well. <futurile>aitzkora: yes - it's the same -L / --loadpath=./ for every command <aitzkora>@futurile, thank : Yes it seems to work, do you know how to load a bunch of commands at the start of the REPL (to avoid to enter ,use (guix) for instance) <Geryz>Hi, so I've been considering switching from NixOS to Guix System (due to several reasons), but since the package repository is a lot smaller, I wanted to check that all my favorite Nix packages are also available in Guix. However whenever I connect to packages.gnj. <Geryz>.gnu.guix.org*, I get 504 - Bad Gateway Error <Geryz>packages.guix.gnu.org* sorry <apteryx>ieure, podiki: got the nss-certs warning produced and automatically handled in (gnu system); would you like to review it? <coyote>is anyone else having problems building an "acl <coyote>is anyone else having problems building an "acl" derivation when reconfiguring their system? <Geryz>So is packages.gnu.guix down/under maintenance or am I doing something wrong? <martin2020>I also haven't been able to access these sites for the past couple days <Geryz>Very strange, hope that gets fixed <martin2020>Is anyone involved with running the web infrastructure here? <Geryz>Also, is there any other way of searching for available packages? <wakyct>maybe install guix in a vm Geryz? Then you can search from there <wakyct>idk if that uses packages.gnu infrastructure though <freakingpenguin>Geryz: Nix has a guix package, right? Maybe you could install it from there, run guix pull to get latest packages, and search with that? <wakyct>aitzkora, perhaps configure the repl in your .guile file? <Geryz>Thanks for all the kind answers <Geryz>Yea I just noticed you can install guix as a nix package, even tho that seems kind of counter-intuitive since they do pretty similar things (shouldn't there be some kind of collision?), that seems pretty reasonable to try it out in the first place <Geryz>dariqq peanuts The hpc index also seems to work pretty well <freakingpenguin>The installed software itself won't conflict, guix runs on "foreign distributions" very well in my experience. (info (guix) Installation) explains, although I don't know how many of those steps Nix takes care of for you. <coyote>NixOS has a Guix service definitions so I assume it does handle some <coyote>much like how Guix System has a Nix service :-P <thanosapollo>I'm finally almost done with my guix setup, is qtile 18.1 broken on guix? Tailscale and qtile are the only ones I havent set up yet <futurile>thanosapollo: have you had a look in the bug tracker to see if anyone else has seen an issue? (issues.guix.gnu.org?) - sorry I don't use qtile <freakingpenguin>I think I heard someone else on IRC mention having trouble with qtile recently <thanosapollo>There aren't any issues open, but it just fails to build due to xcffib conflict, I will investigate this next <apteryx>I no longer seems to get the ack with issue number by email when sending patches to guix-patches@gnu.org; is it just me? <apteryx>20 minutes seems like it should be enough time <freakingpenguin>apteryx: I got an email ack after my patch tuesday, so if debbugs has a problem it started after that point <apteryx>thanks for the confirmation; maybe something with my provider (gmail) or my filters <apteryx>it's been ongoing for a couple days/weeks I think <apteryx>hm, I've got some hang with 'guix system reconfigure' <lilyp>I think shepherd hanging on reconfigure might be related to that other fibers bug. <lilyp>but that's just a hunch of mine <twn>when running guix publish, i'm seeing the following error: <twn>Throw to key `avahi-error' with args `(#<avahi-error-enum Daemon not running> make-client)'. <twn>only when I pass in the --advertise arg <twn>am I missing something simple? I feel like I am <ieure>twn, You're asking it to communicate with Avahi, which you don't have installed. <twn>avahi-utils installed fixed it <Franciman>Hi, I would like to contribute adding a package to guix. It is a rust package, and I have a question <Franciman>i saw that in some packages the check phase of cargo is skipped <Franciman>should i leave it? AFAICT in the check phase you redo the build from scratch and do tests <apteryx>getting a segfault from vlc due to video acceleration on some machine after upgrades; anyone else? <apteryx>it looks a lot like #63197, but ~/.cache/mesa_shader_cache no longer seems to exist <apteryx>Franciman: opinion may differs, but my traditional stance is that tests are useful, even for rust crates <apteryx>I think there's value trying to have them run <Franciman>i will try to post the patch and wait for suggestions then :) <coyote>yeah in my personal opinion tests should only be skipped if they fail upstream or for some other reason <Kolev>Has anybody tried setting up plymouth-encrypt? <apteryx>ieure: the nss graft patch is in #70569 <minikN>freakingpenguin: Thanks for sharing that, it fixed my problem. <ieure>apteryx, Looks good to me, I'll pull out my personal machine and indicate that in the patch when I break for lunch. <Guest92>Is guix not installable at the moment? <attila_lendvai>Guest92, did you try with a freshly generated iso? i know about one issue: grub in current guix is too old to handle the ext4 partition that the current e2fstool generates. <Guest92>yeah i’m in a vm in gnome-boxes. I am trying to reconfigure but can’t because of Nss-certs <Guest92>there are two entries and i guess that doesnt work <podiki>apteryx: i won't be able to test the nss patch, but my quick look at the first patches (moving/grafting) looked good <Guest92>is there an ignore flag or something? <podiki>Guest92: remove "nss-certs" from the system configuration file <Guest9>commenting out the “packages append nss-certs” bits worked but are there any side effects? <sham1>Shouldn't be, since the nss-certs are in the base packages <Guest9>Right now I am in a vm but I want to make an ISO to install on my laptop. Do I just “guix system image config.scm”? <Franciman>when the software i'm packaging requires that i also package various rust dependencies, what should i do? Do i create a single patch? <futurile>Franciman: yes, one crate per patch - remember there's the importer - I didn't realise that the first time and did them all by hand, heh <futurile>Franciman: yeah, welcome to Rust - why have one package when you could use ten ;-) <Franciman>unfortunately i can't escape it lol, texlab works much better for me than texlive-digestif <freakingpenguin>Rust: why have 1 package when you could use 10? Go: Why have 1 package when you could use 100? <dthompson>javascript: why have 1 package when you could use 1000? <freakingpenguin>"Hang on, let me just spin up a headless web browser to generate a flowchart" <decfed>anybody know how to achieve the same as --cap-add=sys_admin when creating a container using guix shell? <decfed>I want to mount a FUSE filesystem inside the container and cannot work out the right incantation of guix shell -C <Kolev>efraim: Do you feel like tackling my audio issue again? <Kolev>By the way, I propose that Guique be the singular of Guix. :P <apteryx>podiki ieure `guix build /gnu/store/p6p103kf2b58zqwkl05mhqs1gd3gc6zr-icecat-115.10.0-guix0-preview1' should give you a substitutable icecat using the grafted nss 3.98 <ieure>apteryx, I've never seen a command like that, where you specify the store path -- is that documented somewhere? <ieure>I applied your patches to master and am building icecat with the pre inst env. <ieure>Sigh, my offload setup is still broken because some *other* code wants a SSH public key for some reason. <sham1>Hmrm, maybe instead of having my service generate xinitrc as a bash script, I could make the service deal with gexps directly and generate a guile script <sham1>A bit annoying, but probably less so than having to deal with trying to wrangle shell lines out of lists of gexp <twn>running into the following issue that i can't explain the behavior and was hoping for some input. tl;dr: trying to build packages locally and serve them via guix publish and sometimes the client is getting errors surrounding a missing derivation on the sub server, however the package is there but i see no derivation in /gnu/store <sham1>Oh well, the shell generating thing works for now <zeropoint>apteryx: I had an issue with vlc just now too, will probably dig into that this weekend <apteryx>zeropoint: it boils down to rm -r ~/.mesa_shader_cache <wakyct>twn I don't understand, is the client even connecting to the sub server? <apteryx>ieure: not sure! it accepts .drv as well <twn>wakyct yes. it connects just fine and i can see the 404s for the narinfo files and and GETs for some substitutes (when things aren't broken) but for some reason (nano in this case) wont download the substitute for nano because of a missing .drv file (from what i can read) on the substitute server <twn>if i supply the --fallback flag, it will continue building locally and i might be okay with that, i'm just not able to make sense of why this feels as though it's in a weird state <twn>or that i'm missing something <apteryx>podiki: icecat with nss 3.98 seems to work normally <wakyct>does looking up the drv for the package return nothing? Or were you visually scanning the store files <ieure>apteryx, podiki, Agree, icecat seems to be fine. I'm building Librewolf now. I saw someone on guix-devel mentioned nss 3.99? ughhh <twn>wakyct i was visually scanning the store files on the filesystem <twn>and you'll have to pardon my ignorance/misunderstanding here but does guix supply a way of pulling/fetching/"looking up" the drv file? or how are you suggesting i look it up? <twn>wakyct oddly enough, my shell timed out and i relogged and tried again and everything is just working again <twn>i'm logging off for a little bit but if you respond, i'll search the irc logs later.. i appreciate the help and will be digging deeper into those links.. again thanks