IRC channel logs
2025-01-08.log
back to list of logs
<RavenJoad>Does Guix have GCC's Ada compiler (gnat) already? I cannot seem to find it and need it to build GHDL. <the_tubular>Can I install both emacs and guile-emacs in the same profile ? How is it gonna work ? <elevenkb>Hello there, is there a way to see why a certain home-configuration object would depend on a certain package? <Kolev>Hi all. Really enjoying my Guix System with Sway. <meaty>elevenkb: if it's a package or service, use guix edit. Otherwise you can also grep through Guix itself <meaty>Is there a let-like form of use-modules? Something that says "use these modules only within this sexp"? <meaty>Also is there a channel for melpa-listed emacs packages? <efraim>can I run the tests in (tests toml) using guix build? I've had success with other tests like that <efraim>looks like I can run them with ./pre-inst-env guile then (use-modules (tests toml)) <efraim>not ideal for testing with offloading <acid-bong>moin, fellers and felines. is there a repo (channel? in guix terms) that automatically syncs Emacs packages from ELPAs (at least gnu, nongnu and melpa) and translates them into guix expressions? akin to Nix's emacs-overlay <efraim>well I guess that's good the toml tests passed <csantosb>RavenJoad: ghdl (and ada) is part of guix-science channel too <jlicht>acidbong: I've assumed from reading the description it's FOSS; please verify for yourself too though <jakef>acidbong: guix import elpa doesn't do the job? <lilyp>running `guix import` every time has its limits – there is a reason to keep a periodically updated channel, that (as far as I'm aware) just periodically runs `guix import` or `guix refresh` <rekado>re matrix-nio: it's fixed on the python-team branch <rekado>I'm rebasing python-team on top of rust-team now; python-team is third in line for merging into master. <efraim>rekado: mesa-updates was merged, rust-team is next up I thought <efraim>ah, there's libgcrypt-gpg-update in between <efraim>I'm hoping to finish building rust-1.54 on berlin before merging rust-team, I think it should be ready later today <efraim>rekado: I think I updated the rust dependencies for all the python packages, so they shouldn't be held back by that <the_tubular>Just to understand, which part is blocking so we can't get a much newer version of rust ? <efraim>we're sticking with 1.82 for this round since I had to rebuild everything for testing after upgrading mrustc. I'm just waiting for the build farm to catch up a bit before merging it <efraim>its around 2000 commits, mostly of crate updates <efraim>ah, we use mrustc to build 1.54, and then we build each newer version in turn <efraim>mrustc targets x86_64 first and then fixes support on other architectures, so it's not quite ready for mrustc->1.74 in guix <efraim>mrustc isn't very fast compared to the version of rust it builds, so the first version takes a while <acidbong>jakef: does it turn elpa into a guix-compatible channel? I'd like my config to be as much declarative as possible <efraim>IIRC it's about 75 minutes on my machine, and then 1.55 takes about 20-25 minutes <efraim>riscv64 takes about 20 hours with the newer mrustc compared with ~28 before <jakef>acidbong: not quite, it spits out a guix package definition for the emacs package, and sometimes it works! <efraim>the build phase for rust-1.54 on berlin took 5109 seconds <efraim>aarch64 is on 228/230 for rustc, then it builds cargo <efraim>oh, I have numbers on my machine <efraim>build phase: x86_64: 3134 seconds, aarch64: 11900 seconds, and I apparently lost the logs for riscv64 <efraim>build phase for rust-1.55: x86_64: 622.7 seconds, riscv64 20887 seconds <rekado>I just rebased python-team on top of rust-team; now I'm getting "Error: Could not set character size" on "make doc/images/service-graph.png". This looks familiar... <efraim>I'm running out of things to disable in the 'build phase of the pyproject-build-system to see what's broken on powerpc-linux <efraim>ok, I assume it's not (file-exists? "pyproject.toml") that's causing the crash, so it's probably parse-toml-file <rekado>efraim: we recently applied a change to the toml parser on python-team <efraim>I'd have to check my calendar and the git logs to see how far back it might've been. my powerpc machine was offline for a while <efraim>i'm getting "build process 3 exited with status 11" on python-tomli on powerpc-linux <efraim>(guix build toml) is self contained, so I can load that by itself on the powerpc machine without needing to compile everything and test it out if necessary <efraim>reverting the change from last may didn't make a difference <f1refly>I packaged the varlink c library (https://varlink.org/) so I could build kanshictl, into what package file should the libvarlink package definition go? <efraim>parse-toml-file doesn't choke on tomli's pyproject.toml, so that's probably a good sign <efraim>rekado: could it be my guix-daemon is too old? <f1refly>I suppose it could go in serialization.scm? <lilyp>I take it the `git send-email` workflow doesn't quite work for you? <Deltafire>rovanion: did you read the rest of your thread? looks like it's been merged already <lilyp>meh, forgot to edit the commit message of musescore for v2 <cluelessguixer>Is it just me, or does guix not play well with tramp for remote editing with emacs? Doing C-x C-f, typing in "/ssh:server:" and trying to tab complete yields nothing. Seems to work on other servers. <lilyp>do you use plain old emacs or emacs-tramp? <lilyp>it should work with plain old emacs but I don't remember whether emacs-tramp is broken in that regard <rovanion>lilyp: It is so seldom that I contribute to anything that uses `git send-email` that I always have a broken setup when I try to use it, so I've given up on it. <cluelessguixer_>lilyp: Should just be plain old emacs, but it's not on a guix system (the remote system is though). <lilyp>hmm, that adds a possible layer of breakage <lilyp>if you have guix on your foreign system, try running emacs from there (using a guix shell if you don't want to mess with your system's emacs) <dariqq>cluelessguixer_ : Try adding 'tramp-own-remote-path to 'tramp-remote-path <cluelessguixer_>dariqq: Seems those variables are void. Only thing I see in customize-set-variable is "tramp-ignored-file-name-regexp" and "tramp-mode". <dariqq>(or probably better: only adding /run/current-system/profile/bin instead of the whole PATH) <cluelessguixer_>I appreciate the help, dariqq. Unfortunately I don't have those tramp variables. It should be built in and not a separate package, right? <dariqq>cluelessguixer_: hmm , not sure what is going on. I have it on my emacs (but i am currently dont have a guix machine near me to check there) <dckc-e>i just installed guix on this machine (thinker, a thinkpad). picked exwm for the UI. M-x guix didn't do anything. <dckc-e>I did M-! guix install emacs-guix , but M-x still doesn't do anything. How do I get emacs to notice that I installed emacs-guix??? <dckc-e>I guess I'll restart exwm and restart IRC and such. If there's a better way, I'd like to know about it, though. <dckc-e>the stuff I was just talking about is gone... I wonder whether emacs irc keeps logs <dckc-e>hm. M-x guix still fails silently. C-h f guix shows nothing. <futurile>sorry dckc-e I guess you're going to have to wait for someone that uses emacs to answer. It's pretty quiet around here at the second ;-) <dariqq>dckc-e: youd need to also install emacs into the same profile to get the EMACSLOADPATH updated as well. (See the Search Paths section in the manual for more details) <cluelessguixer_>Managed to find tramp-remote-path and give it an entry of "/run/current-system/profile/bin", but it still doesn't work. I can do "/ssh:server:test" and somehow open the editor (looked to open in ~/), but trying to open anything using absolute paths just fails with "no proper ls command". <RavenJoad>Does Guix have gnat (GCC's Ada compiler) packaged somewhere? I went looking and didn't find anything. <Kolev>RavenJoad, I'm not finding anything either. Also, "ada" and "gnat" are actually quite difficult to search for. <RavenJoad>Welp. I guess I'm about to go on an adventure then. <Kolev>Oh, wow. Why isn't it in `guix`? <RavenJoad>That's a solid stop-gap. I'd like to upstream Ada & gnat to Guix though. I mean, it's part of GCC. <Altadil>RavenJoad: Not yet. Currently, there is no way to bootstrap it. <RavenJoad>Hmm... Perhaps that will be my next project, once I get all my other outstanding Guix patches pulled in. <cluelessguixer_>Did I get it right? "(sudoers-file (plain-file "sudoers" "host ALL=(ALL) /home/host/.guix-home/profile/sbin/nvme smart-log /dev/nvme0")" <dthompson>I'm looking for a committer/maintainer to help get jlicht's node 22.12.0 upgrade merged. it affects enough other packages that we really need to pre-build everything somewhere before pushing the commits to master. <mrvdb>ah, missed that. GOOPS didn't ring a bell with me <cluelessguixer_>I locked myself out of running "sudo -E guix system reconfigure ~/.config/guix/system.scm" now, lol. <dariqq>cluelessguixer: If that does not work you can try booting the previous generation in the grub menu which should have the correct sudoers file <cluelessguixer>I guess a different name to plain-file doesn't do it either. Time to reboot again... <dhoffman>ekaits and/or efraim: are the risc-v `make check` tests supposed to take a long time to finish? i have the wip-riscv-bootstrap branch checked out and followed the building from git instructions up to make check (building in the guix shell environment provided as an example). just wanting to make sure this is expected behavior <cluelessguixer>Alright, I figured out how to edit sudoers now. Just need to cat what's there already and then add what you need. <cluelessguixer>Was hoping "host ALL=(ALL) NOPASSWD: /sbin/nvme smart-log /dev/nvme0" would allow me to run it without entering sudo password, but it seems not. <RavenJoad>cluelessguixer: You may want to look at the privileged-program system service-type. That would let you put that program into a setuid/setgid place, without going through and editing /etc/sudoers manually. <homo_>cluelessguixer guix has doas (package's name opendoas), its config syntax is much easier to understand: https://man.openbsd.org/doas.conf , my approximation to what you want is: "permit nopass : cmd /sbin/nvme args smart-log /dev/nvme0" <muradm>could some one look at 75270? thanks in advance <Epsilon13>hi there, i m trying to setup an emacs with eglot environment on guix for rust <Epsilon13>but eglot return me a Failed spawning proc-macro server for workspace `/home/epsilon/projects/hello_world/Cargo.toml`: cannot find proc-macro server in sysroot `/gnu/store/xj7fd4v\ <Epsilon13>3j368jins9lvld09n4vs2i23p-rust-1.82.0` when launching eglot inside a rust file <Epsilon13>if anyone could help me setting up a rust environment for emacs in guix that would be delightful <cluelessguixer>Looks like "(kernel-loadable-modules (list (list zfs "module")))" did something. <ekaitz>hi! which texlive packages provides \usepackage{tikz}? <dhoffman>ekaitz: misspelled your name with a question about the risc-v enablement effort, not sure if you caught that <ekaitz>probably not, tell me again please <dhoffman>are the risc-v `make check` tests supposed to take a long time to finish? i have the wip-riscv-bootstrap branch checked out and followed the building from git instructions up <dhoffman>to make check (building in the guix shell environment provided as an example). just wanting to make sure this is expected behavior <ekaitz>dhoffman: i need more specific info <dhoffman>this is the long-running command (everything other qemu-binfmt process is a child of this) <dhoffman>/usr/libexec/qemu-binfmt/riscv64-binfmt-P /gnu/store/iaf2d97db3h6mav467i9v3999ab8f9sb-guile-bootstrap-2.0/bin/.guile-real /gnu/store/iaf2d97db3h6mav467i9v3999ab8f9sb-guile-bootstrap-2.0/bin/guile --no-auto-compile -L /gnu/store/6qhr7s8431yv7ikcax3l30hw6cim95kc-module-import -C /gnu/store/k38byh45jnjph6jlxb6lmrrqbvij1qfq-module-import-compiled <dhoffman>/gnu/store/hrf4zpl5raldb8vqw8yj0f743vyhdxlx-make-mesboot0-3.82-builder <dhoffman>maybe i can re-run the make check with some verbose prints so i can see which ones begin but don't end (within a reasonable time frame) <ekaitz>dhoffman: how long is it running already? is it using your cpu? <dhoffman>there's a consistent 15% CPU usage (for context I have an i9-13900K). it's been running for about twelve hours at this point <dhoffman>i understand the full bootstrap process can be slow but i'm not sure whether the `make check` tests do anything that heavy <ekaitz>12 hours seems a lot, but it might be legit <ekaitz>we have had some hangs but those made cpu usage go to 0 <dhoffman>are there any valuable debug outputs in the tests/* subdirectory that I can search for? <dhoffman>looks like it only prints pass/fail and not when a test starts <ieure>If you give your operating-system definition to that function, it'll add ZFS stuff to it. <ekaitz>dhoffman: i would just wait for some hours more and just in case ask efraim <dhoffman>i'm busy until the end of the workday anyways so waiting isn't a problem <ekaitz>efraim has more experience than me with the current setup <ieure>cluelessguixer, ZFS in Guix is still pretty basic at the moment, it needs a good amount of work still. <cluelessguixer>Heh, just imported my zfs pool (forgot to export it, but no problems). What a beautiful sight. Will I have to import it on every boot or is that not the case anymore? <cluelessguixer>kernel-module-loader-service-type. Don't think I needed that. Can't remember if I did modprobe after rebooting. <ekaitz>civodul: is there any chance that packages that are in texlive are not available as standalone? <ekaitz>i'm thinking about `calc`, i don't find any texlive-calc package <civodul>ekaitz: yeah i think Nicolas imported the whole thing <ekaitz>civodul: yeah, thanks! they are a little bit hard to find sometimes... also it's hard to know which ones are the minimum set i need for my goals... <ekaitz>also i don't understand very well what's the best way to add them as dependencies... inputs vs propagated inputs and if will be able to add stuff ad-hoc and if it will be found... <ekaitz>maybe it's just easier to install the full texlive... <rekado>ekaitz: there are "collection" packages, too <ekaitz>i'm trying with texlive-scheme-small now <RavenJoad>I have had an issue open for 6 months now (72249), and have made improvements. I also want to add some other related stuff to that patch train. Would it be best to -v2 it? Or just open a new issue? <ekaitz>RavenJoad: -v2 i think it's better <RavenJoad>ekaitz: Even if there is more stuff? The original only added 1 service-type, the v2 adds 2 mutually-exclusive sibling services. <ekaitz>RavenJoad: is there the possibility of finish the first one and add the other service in a different one? <RavenJoad>Yep! They are independent of one another, but have the same provision symbol. Both are guest agents for running Guix in a Xen VM. <RavenJoad>Excellent. I have 72249 and 75445 sitting, waiting for review now! Hopefully Guix guest VMs can report back to their Xen hosts! <RavenJoad>Both patch trains are working perfectly for me on my XCP-ng setup right now, so I'm hoping this can just go through. <lockbox____>RavenJoad: nice! Looking forward to being able to use xen-guest-agent, I recently (monday) was looking to see if someone had already done that :D <stochastic>> Excellent. I have 72249 and 75445 sitting, waiting for review now! Hopefully Guix guest VMs can report back to their Xen hosts! <clone>is there a way to see when a package stopped building?