IRC channel logs
2025-08-22.log
back to list of logs
<apteryx>on the very first request in many days. <corvvs>Hello, do I have to use the `-E` flag in this command? sudo -E guix system reconfigure ~/.config/guix/system.scm; or can I drop that -E flag? <podiki>you don't need it on guix system <podiki>(which the only place you'd be running that command, but anyway) <podiki>the only command you really ever need it for on guix system <podiki>(guix system build doesn't need sudo, but to write the bootloader you need it; that's part of what reconfigure does) <corvvs>I'm following systemcrafters' 2021 tutorial lol <podiki>that's a bit old (probably nothing changed for the basics other than minor stuff) and third party so to speak, so you may need to ask them directly for any particulars if it is their installation image <corvvs>Now I have a problem with the guix system reconfigure command lol. It said "ice-9/boot-9.scm:1685:16: In procedure raise-exception <newline> Git error: SSL error: received early EOF <corvvs>It's happening many times and usually happens at "receiving objects 20% to 40%" <corvvs>Why is there a /run/systemd folder? <corvvs>Btw, when installing emacs with "guix install emacs" why does guix pull in many irrelevant stuff? Doesn't the base system already have those? <lilyp>it's specifically libgccjit, which IIUC is not on most systems <lilyp>other than that, there's probably a number of inputs specifically to emacs <lilyp>plus, if you build from source and don't typically do so, that's when you'd fetch gcc <nikolar>I don't think you need GCC installed to use libgccjit <csantosb>Funny: `guix graph --path emacs binutils` but `guix graph --path emacs libgccjit` <corvvs>I literally can't `guix system reconfigure` at all lol... It's been almost 5 hours since my first attempt and all 7 attempts to run the command failed at around 20% receiving objects with SSL error <corvvs>My internet is fine, heck, I can stream 4k YT vids just fine <andreas-e>csantosb: Yes, somehow the data service does not evaluate it: <andreas-e>The reason is unclear, since other branches do work. I have moved emacs-team one position back in the queue waiting for the problem to be resolved. <andreas-e>corvvs: Is it always the same object? Because otherwise you could continue - what has been downloaded into the store, stays in the store; so there would be progress. <Rutherther>'receiving objects' is just download from channels repositories, there is nothing being put to the store, only to root's cache <csantosb>andreas-e: do you think this is something which might be reproduced locally to debug ? <corvvs>Rutherther: so what should I do then to resolve this problem? <corvvs>This person had the same problem as mine, but no one responded lol <Rutherther>one thing you could do is get around this by using sudo -E to use your user's channel cache <hugohugo>Guix keeps a cache of all git clones. I had some similar problems and had to delete the right cache. See .cache/guix/checkouts <Rutherther>but the issue is not with the cache, the issue is with connection to channel's repo <hugohugo>Yeah but still something to try. Maybe making a fresh clone would somehow work differently from updating an existing one. I don't know <corvvs>:facepalm: Adding the -E flag solved the issue... <corvvs>Thx guys, sorry for wasting your time (sigh) <Rutherther>there is nothing to facepalm, this is not a solution to the issue, it is just a workaround, and a bad one <andreas-e>civodul: Yes, I have been on it since the start; it is a PR that goes with the issue 2139, on which I had already reacted. This is the third insight-toolkit PR that I am shepherding through... <aeolem>hey, i'm trying to install (non)guix on my old laptop again (i've done it before) but i keep getting this strange "guix system: error: writing to file: broken pipe" error when trying to run "guix system init" <aeolem>i mean i have invoked the command before which errored out due to syntax errors, but that shouldn't do anything, right? or did it somehow corrupt my /mnt <aeolem>ah, found the mistake, i forgot the "https://" in front of one of the substitute URLs <aeolem>now why that results in a "broken pipe" eror and not a "error resolving substitutes" error is beyond me, but fair enough <jakef>andreas-e: not sure if you noticed, but it's the Python wrapping that makes ITK take so long to build... insight-toolkit-legacy builds in a flash, relatively <andreas-e>jakef: I think I had noticed it the last time, yes. <aitzkora>Hi, which function could I use to create a temporary dir (for some python tests I need to do that) <andreas-e>With the two patches together things are fine, I am just letting everything build once now. <andreas-e>aitzkora: Inside a package build, you mean? Maybe just "mkdir" then? Since anyway the build directory will be thrown away in the end. <aitzkora>andreas-e: It could be work, but due to python path I want to run in another directory to make the tests more isolated, I'm not sure that in a sub-directory it will be solve my troubles <andreas-e>In any case you cannot escape the build directory. So by definition it will always be in there. <jakef>I've been trying to package EGSnrc. It uses Mortran, which has an 80 character file path limit... I'm now stumped <jakef>I might submit what I have so far as a WIP PR and hope for some help :) <ajarara>hello #guix, (shepherd service timers) requires guile-fibers (which makes sense, given the requirement these jobs be interruptible and run on the shepherd thread so a lock up is real bad). Doing something like 'guix shell guile-fibers' doesn't quite work. What am I missing here? <ajarara>I'm following the info pages for Shepherd, 4.8 <Rutherther>ajarara: you're missing the package that provides the necessary search path in your shell - guile <Gooberpatrol66>my ~/.config/guix/channels.scm is generated using the guix home service <Gooberpatrol66>is the channel url going to change to codeberg automatically? and when is that going to happen? <Rutherther>Gooberpatrol66: if you mean you're using %default-channels, then with latest guix you should already have codeberg url (or rather the url that redirects to codeberg - git.guix.gnu.org) <Gooberpatrol66>Rutherther: it's still pulling from git.guix.gnu.org for some reason <Rutherther>yes, that is correct, that is the new url that redirects to codeberg since the time the migration happened <Gooberpatrol66>i have a couple other channels but the guix channel is just the default one <jacobk>Should I open a issue on Codeberg?