IRC channel logs

2024-10-01.log

back to list of logs

<nckx>bavier: I'd mention, at most, that it ‘provides udev rules for use with @code{udev-service-type}’ or so, worded appropriately by how necessary they are (if at all). No code snippets.
<bavier>nckx, that sounds reasonable. I think I can work that in.
<bavier>I see a couple other packages, e.g. in (gnu packages radio), do this
<bavier>oh, but they include code...
<elpogo>efraim: I know you worked on minimal guix system images previously so i thought i'd tell you about something i found amusing. A Guix image with i3wm xterm librewolf (and xorg ofcourse) is about 5.2G, whereas building with alpine-make-vm-image with the same packages (s/librewolf/firefox-esr) gives an image just a little shy of 500M - less than one-tenth the size!
<vagrantc>guix has always been disk intensive
<elpogo>And the guix image did not even include guix itself - i followed the tip from efraim's video and deleted guix-service-type from the %base-services
<elpogo>I wonder how much fat I can trim if I supply my own initializer for the root partition remove the info and man pages etc
<aarcov>Is there a way to prune a built guix package down so that it removes the sources and native-inputs (and any other equivalents for other build-systems (ex. 'cargo-development-inputs' when using rust)? That might help to shrink an image somewhat
<stochastic>Rutherther, thank you again, it works very well
<elevenkb>Hello there, there's an incompatibility between texmacs and hunspell. The Hunspell British English dictionary uses the en_GB-ise.aff but texmacs is expecting en_GB.aff. I'm wondering if there's a quick way to patch the hunspell dictionary package so that it adds an en_GB.aff file?
<cpli>should you ever `guix pull`?
<cpli>or in another way of saying it: "how do i update my package index?"
<bavier>cpli, `guix pull` is the way to do that.
<cpli>oh alright, if i want to update nyacc, to the latest blob available: https://download.savannah.gnu.org/releases/nyacc/nyacc-2.00.5.tar.gz
<bavier>cpli, that's a bit different. For that, you'd be updating the package definition. There's also `guix build --with-tarball=...`. See the manual for those options.
<cpli>bavier: should it be sufficient to change the (source (origin ...) in gnu/packages/mes.scm via guix edi-
<cpli>oh, alright i can do that too
<cpli>thank you!
<bavier>cpli, np, cheers.
<cpli>bavier: i believe you mean `--with-source`, but i found it
<cpli>bavier, actually after building successfully, how do i "install" the resulting storepath into my profile?
<afm-victoria>Is gnu.org down?
<afm-victoria>I am having issues with git.savannah.gnu.org
<AwesomeAdam54321>afm-victoria: Yes
<afm-victoria>OK.. thanks.
<AwesomeAdam54321> https://hostux.social/@fsfstatus/113230041959498522
<dcunit3d>is savannah.gnu.org literally in savannah, ga?
<afm-victoria>AwesomeAdam54321: thanks for the update. I borked a system update but I was able to rollback and get back to a working system. Win for Guix!
<afm-victoria>Elpa must be having issues too
<apteryx>is documentation considered 'non-functional' data per the GNU FSDG?
<apteryx>I'm trying to package https://github.com/libretro/slang-shaders/, and it's a mess because it's a huge collection of random projects, some free, some unknown, some not. So far I've made an allow-list preserving only the fils I've explicitly told it to in the source snippet; sometimes these are configuration files or png or READMEs without any license attached... wondering if these use/relate to an
<apteryx>underlying free shader source, is it reasonable to include them?
<dcunit3d>apteryx: hmmm i wonder if you can trace back some of the files to their original introduction, though that's unlikely to be efficient
<apteryx>oof: https://github.com/libretro/slang-shaders/pull/641
<dcunit3d>it seems like a cool project though.
<apteryx>dcunit3d: yeah I'm not prepare to waste weeks on this
<dcunit3d>i see, some of these are loosely based on shadertoys? maybe the maintainer can clear it up
<apteryx>lots of them originate from shadertoys, and the default license for shadertoys unless explicitly mentioned is a non-commercial Creative Commons license
<apteryx>(which is thus non-free)
<apteryx>wonder if such 'trivial' code is covered by copyright or not: https://github.com/libretro/slang-shaders/blob/master/include/compat_macros.inc
<apteryx>a bunch of defines adapting for name changes across API version
<apteryx>I guess it's copyrightable... copyright protect the form/shape, not the ideas
<adanska>has anyone heard anything from the savannah sysadmins for getting the guix repo back up?
<adanska>oh... www.gnu.org is down too..
<futurile>the hostux link awesomeadam gave says they don't expect anything until 'tomorrow' - depending on what timezone one is in I guess
<adanska>yeah just read that. hopefully it gets fixed soon...
<janneke>futurile: there are time-zones outside of the world^W^W america?
<erin93>Hi, I'm trying to run the android command line build tools in a fhs container and getting a "sh ... no such file or directory" error. Does anyone know if there is some package I am missing or something I need to do differently? thanks
<stochastic>> oh alright, if i want to update nyacc, to the latest blob available: https://download.savannah.gnu.org/releases/nyacc/nyacc-2.00.5.tar.gz
<stochastic>if you want the latest, you can literally say `--with-latest=...`
<Rutherther>sneek later tell erin93: so you are trying to run a precompiled binary, right? Can you share the whole error including the file that is not found? And in relation to that, what is the name of the executable you are running?
<Rutherther>Hmm, what have I done wrong that sneek is not answering? The colon?
<Rutherther>sneek later tell erin93 so you are trying to run a precompiled binary, right? Can you share the whole error including the file that is not found? And in relation to that, what is the name of the executable you are running?
<Rutherther>Is the bot dead? Hosted somewhere under gnu that is down?
<futurile>janneke: I believe there's just 'East Coast', 'West Coast' and ... 'WTF!?' .. heh
<nckx>sneek: botsnack
<sneek>:)
<nckx>sneek: tell Rutherther to indeed check their colon.
<sneek>Rutherther, nckx says: to indeed check their colon.
<nckx>Rutherther: Nah, it's not an official GNU bot, you can safely say 'Linux' in its presence. It's hosted and maintained by a #guile regular.
<Rutherther>nckx: oh, colon after the sneek's name. Well lets try for third time.
<Rutherther>sneek: tell erin93 so you are trying to run a precompiled binary, right? Can you share the whole error including the file that is not found? And in relation to that, what is the name of the executable you are running?
<Rutherther>Okay, it seems I am a lost cause :)
<nckx>Sneek isn't very bright. "Tell" without explicit "later" means right now, like I did above. I presume it's not working because erin93 isn't here, but I don't know whether it falls back to "later" in that case.
<bost>Hi I can't `guix pull` from https://git.savannah.gnu.org/git/guix.git am I the only one?
<nckx>Savannah's down.
<jakef>poor savannah.gnu.org is down?
<nckx>We're not allowed to lie in the channel topic. https://hostux.social/@fsfstatus/113230041959498522
<nckx>(I'd add that link but I'm 97% sure that it would truncate the topic.)
<amano>How can I pass --allow-discards option to luks mapped device?
<amano>It seems guix doesn't currently support this.
<amano>allow-discards is pretty necessary for SSDs.
<amano>"cryptsetup --allow-discards --persistent open /dev/sdaX root" saves --allow-discards option in LUKS2 metadata.
<nckx>This sounds familiar, I wonder why I'm not finding any search results (besides the obvious answer).
<nckx>Anyway I didn't think it was supported.
<futurile>efraim: do you push your Rust-branch WIP anywhere accessible?
<futurile>efraim: obviously today is the *day* I would have planned working on a Rust package heh
<jakef>thanks nckx i didn't notice the channel topic
<jakef>just got a guix pull through
<nckx>Yup, I just came in to change it back :-)
<dariqq>regarding my test failures on i686 from yesterday: Skia and node failure look like integer overflows/ errors becasue number too big
<efraim>futurile: its on savannah as rust-team, and I also mirror it at https://git.sr.ht/~efraim/guix
<futurile>efraim: thanks
<gnucode>hey guix, I'm thinking about trying to buy an open source router supporting 1GB speed. Is there anything better than turris.com ?
<mirai_>gnucode: I'd ask #openwrt @ OFTC
<gnucode>mirai_: yeah I can't seem to use their webchat at the moment...
<bigbookofbug>hi all, im having an issue with guix authentication
<bigbookofbug>so i have a channel that used an old key, which i no longer have access to, and i am unable to push with my new key
<bigbookofbug>i get the following error: guix git: error: initial commit 05a6e2e95e4d7ed26afe19ddd9782aa972536fae is signed by 'ACF7 CA25 C886 9B8A 1669 67D7 7350 3372 E2C6 3BCF' instead of 'EF51 5F7D 6007 1778 1DF9 AB2E 0FB1 CF28 67A1 17F5'
<bigbookofbug>is there a way to bypass this, beyond passing a --no-verify flag?
<bigbookofbug>im reading the docs, but can't seem to find a way to get rid of this old key via guix git authenticate
<duncan>I need to get dig, nslookup &c. These seem to be in bind:utils. I can do guix install bind:utils but I want to specify bind:utils in my config.scm, but it's not happy.
<duncan>I found this web-page, but I don't know what the equivalent scheme functions are.https://guix.gnu.org/manual/en/html_node/Packages-with-Multiple-Outputs.html
<duncan>Can anyone help?
<ekaitz>duncan: (list bind utils)
<ekaitz> https://guix.gnu.org/manual/devel/en/guix.html#Globally_002dVisible-Packages
<nckx>Correxion: (list bind "utils")
<nckx>Often written as `(,bind "utils"). Both are completely equivalent.
<duncan>neat, thanks - I have (list (specification->package "bind") "utils") which seems to work nicely
<nckx>It's very roundabout but it's not wrong.
<graywolf>sneek: Who you belong to?
<nckx>duncan: If you want to use a list of strings as package list, I recommend using something like (packages (map (compose list specification->package+output) (list "foo" "bar" "bar:buzz" …))) instead.
<nckx>graywolf: dsmith, not in this channel.
<graywolf>Ack, thanks :)
<duncan>ah, I get it now, the atoms from use-package-modules are equivalent?
<nckx>Uhm, I don't think so, but I'm not sure what you're asking. (use-package-modules a b c) is mere syntactic sugar for (use-modules (gnu packages a) (gnu packages b) (gnu packages c)) — phew, you see why there's sugar. I'm not sure how that relates to listing packages?
<nckx>Most people use ‘specs’ (=strings like "bind" or "bind:utils") because these work without importing each required package module, which is nice, but I don't see how the two are related.
<nckx>s/Most people use …/Most people who use … do so because/
<nckx>ACTION 😴💤