IRC channel logs
2026-05-10.log
back to list of logs
<apteryx>there's this (letsencrypt-certificate "webssh.example.com") used in the manual, but letsencrypt-certificate doesn't appear to be a thing <apteryx>I thought it was some helper to form the expected file name of the certbot cert for a given domain. <adanska>I'm trying to figure out how I can look at the lowered version of a shepherd service as it appears in an operating system record. I seem to be having issues with having a list of lists of strings being perhaps incorrectly lowered in the shepherd service? <adanska>If you look at the end of the paste, I've included the nebulous error that the shepherd is throwing. The strange thing is, when I use (fold-services ... #:target-type network-manager-extra-configuration-service-type) and get the service, then manually lower the start gexp and run it in the store, I get the correct inputs inside the resulting sexp. <csantosb>Wait a second ... "/gnu/store/.../share/emacs/30.2/lisp/treesit.el.gz" is decrypted by emacs on the fly with guix on a foreign distribution <apteryx>adanska: looks like a quoting issue; where lowering gexp one level of quote is lost <apteryx>if you use (list "nmcli" ...) instead of `("nmcli" ...) I think it may be fine <adanska>apteryx, that sounds absolutely correct <adanska>apteryx, thanks a ton, I'll give that a go :)) <theesm_>hi guix o/ working on my reka pr again (https://codeberg.org/guix/guix/pulls/8049), building the rust module by calling cargo-build-system phases inside modify-phases of the emacs package is probably the best shot at avoiding having to have two packages, no? or is there a better way i don't know of yet to combine two build systems <adanska>theesm_, yeah im pretty sure that would be the best way! i think theres an example of mixing in the cargo-build-system into another build system in the cookbook <theesm_>adanska: thx, already have a somewhat working approach i'll push later on (will compare it to the cookbook example, didn't know of it) ... the approach just seemed a bit unergonomic as i'd assume that the number of software having rust parts will increase over time <adanska>theesm_, I guess it's unergonomic but honestly looking at the example in the cookbook it's not too insane. At least i dont think the shape of the phases will change that much, so the same pattern can be reused in most places. <trev>why does this keep happening?: guix system: error: path `/gnu/store/yw1vfd8x1f5pqqmvc27rxqgrdpf44179-bash52-019' is not valid <trev>i did a full gc and verify=repair <adanska>apteryx, I tried you approach but changing the function to use (list ...) didn't seem to change the derivation... there wasnt a rebuild of that service .go file. I'm doing a gc and rebuild just to rule out any problems with the change detection? but odd. <apteryx>ah, that makes sense. Try '#$config-commands <adanska>trev, try `guix gc -D /gnu/store/yw1vfd8x1f5pqqmvc27rxqgrdpf44179-bash52-019` <adanska>apteryx, will do. Why does it make sense if you dont mind my asking? <adanska>im trying to get a better intuition for gexps <apteryx>it's still the quoting issue; you're computing a list before passing it down in the gexp; so it'll look like ("nmcli" ...) in the resulting file <apteryx>but you want '("nmcli" ...), so you need to re-add the stripped level of quote in the gexp <adanska>apteryx, aaaaaahhh i see! that makes sense. this quoting stripping is a good thing to know, would totally tripped me up somewhere else down the line. <adanska>`guix build /gnu/store/yw1vfd8x1f5pqqmvc27rxqgrdpf44179-bash52-019` maybe trev? <trev>just the one that i sowed <adanska>trev, did u try and do `sudo guix gc --verify=contents,repair`? <trev>i'm wondering if i just keep catching bad commits. attempting to make a container with time-machine on a "stable" 1.5 commit <trev>if i directly build --repair the path then it works, but now i have to keep doing it a ton of times <adanska>i mean, bad commits shouldnt violate the store like that. could be some hardware issue? or an improper shutdown? <trev>maybe i forgot to reboot after a reconfigure or something <adanska>reboot probably wont help. I had a similar problem and what i had to do was go and do the `guix build --repair` on all the offending files. it took a while :( <gordon1>looking at the code, i found some patch mechanism that can apply patches to guix channel (commit 053b10c3ef2df7ea80556ab9b2c93d0bf88094f2), it is not documented anywhere, how can I apply a patch to the guix channel in my config? <trev>adanska: at 150 repairs now <PotentialUser-44>hello, i'm trying to use virt-manager and have booted a vm. Now when i want to redirect a usb device to the machine it tells me /run/privileged/bin/spice-client-glib-usb-acl-helper No such file or directory <trev>boxes is lacking features <PotentialUser-44>i've never used gnome-boxes, perhaps it's also good. does it work with libvirtd as well? <PotentialUser-44>In the code of guix, I found mentions of spice-gtk which seems to contain the spice-client-glib-usb-acl-helper. Not sure how to add that globally though. The code I found is for GNOME and I'm on plasma. <futurile>heya all, hope you're having a good weekend <janneke>futurile: thanks, that's still somewhat undecided for me ;) <janneke>hoping you're having a grand weekend ;) <futurile>janneke: I managed to go outside for a bit, just before the rain has set in, so that always makes for a "grand weekend" heh <janneke>ACTION has their 64-bit hurd system booting again, it might just turn out to be a grand weekend for me too <janneke>an `obviously correct` fix turned out to break things <janneke>tusharhero-xmpp: yes, just booted the 64-dbit hurd on an x220 this weekend <janneke>no networking with the 64-bit hurd right now, though... <bjc>from what i can tell your device should work <bjc>i'd guess something didn't get hooked up right in the rumpnet port <janneke>quite possibly, we got to find the right "curses" to get it to work... <bjc>i still haven't been able to figure out how to get a hurd dev environment up in guix <bjc>yeah, i was doing that, too <tusharhero-xmpp>I tried having a childhurd but I can't figure out how to login into it <bjc>i understand the process, it should work. it's basically identical to netdde <janneke>not sure if netdde might be causing problems, for example <bjc>tusharhero-xmpp: ssh -l guix childhurd <bjc>i think the next step for me is to actually compile rumpnet myself, but i'm so far from that <bjc>tusharhero-xmpp: should be able to âvncviewer localhost <janneke>afaik, our "rumpkernel" package just compiles rumpnet in <janneke>and our hurd package just picks that up <bjc>i mean i need to be able to edit source, compile it, and run it in a quick loop <bjc>i think i have all the parts, i just don't know how to put them together =) <stephen0>hello, all. new guix user here. Can anyone recommend a cli tool to format guile scheme? <cbaines>stephen0, emacs, at least that's what I use <stephen0>hmm. Is there a way to call it such that it just formats the file, saves, and exits without having to interact with it? <PotentialUser-44>@noe: thank you, that brings me closer. spice-client-glib-usb-acl- is now found and suid. next step is to get org.spice-space.lowlevelusbaccess registered. <bjc>stephen0: if this is for guix itself, there's âguix formatâ. i don't know of a generic guile tool. i'm sure there's an emacs incantation, but don't know what it is <stephen0>bjc: I dont see that in the help and my naive attempt to invoke it gives an error. <futurile>stephen0: are you reformatting your own code, or something in Guix? (e.g. a package definition) <stephen0>indent level in my sytem and home configurations were weird, guix style -f fixed it. <futurile>futurile: OK, the 'guix style' command has a bunch of rules for guix packages. Be interesting to see how general it is <PotentialUser-44>noe after a reboot, the vm in virt-manager can now see my librem 5 via usb <untrusem>PotentialUser-44: do share your config for future purposes <yelninei>ACTION has too many core-packages-team branches <test202020>some times ago i try to setup autostart for sway, but that fail. <yelninei>janneke: do you know why for coreutils-mesboot configure detects avx2 support for gcc-mesboot but then fails to compile with avx2? <janneke>yelninei: no idea, even what avx2 is <csantosb>Do we have `readelf` ? Diffoscope complains about "'readelf' not available in path. Falling back to binary comparison. " <yelninei>janneke: It detects avx2 intrinsics and then tries to build wc with vectorisation. The assembler does not like something about it so it fails: suffix or operands invalid for `vpcmpeqb' <janneke>yelninei: any idea if hurd-team (your gcc-11 dmd branch) has been built? <janneke>it had too many rebuilds, so i pushed your pr to hurd-team hoping it would get build and we could merge <janneke>yelninei: also, i'm pretty flabbergasted about fsck <janneke>sutils/fsck.c lists --yes, -y, and --force, -f <janneke>no idea where the unknown option errors comes from <janneke>ACTION goes to try --preen --writatable, just like debian has... <yelninei>janneke: i dont think you pushed it anywhere, at least i cant find it anywhere <yelninei>havent heard anything from the gdc/dmd maintainers recently. The last question they had was about ci but theyd need to get a bootstrap compiler from somewhere. Ill ask tomorow if there is more that Id need to do <yelninei>the joys of porting a selfhosted compiler to a new platform <kratacoa>ugh, got disconnected, in case anyone replied in the meanwhile <kratacoa>anyhow, it's probably because my previous generation was just 30 hours ago; I thought it was older heh <bjc>hurd-team has a fix for fsck? <bjc>i don't see any commits since 2 months ago <janneke>but istm (see bug) that our script doesn't even run when root is dirty <bjc>something runs, because fsck complains about arguments <bjc>ah i see. very weird <bjc>i was hoping this'd be more straight-forward <janneke>no idea what's going on here, of course our whole init is completely different from debian <bjc>yeah, i wonder what fsck is being called where <bjc>is there some way in codeberg to watch specific issues without commenting in them? <bjc>all i had to do was ask out loud =) <bjc>(there's a subscribe button on the right-hand side under ânotificationsâ) <bjc>janneke: how hard would it be to change âfsckâ to âe2fsckâ from e2fsprogs? <bjc>that's working for me from the command line, where the fsck that's packged with hurd doesn't <janneke>bjc: that would be quite easy; add e2fsprogs to %base-packages/hurd and call e2fsck <bjc>i mean to replace it in the boot script <janneke>the problem seems to be, that at the moment, fsck isn't getting called when the root fs is dirty <bjc>fsck doesn't work even when it is called <bjc>but yes, two problems =) <janneke>but anyway, good suggestion to keep in mind <bjc>for whatever reason ours only complains about flags <yelninei>the flag it is complaing about is the "/" or any device/fsys option <bjc>yeah, nothing is valid for it <bjc>at least nothing i've been able to summon up <janneke>possibly the --preen without anything works, as soon as we have fstab <janneke>which should be after the first boot <janneke>but we must find out why `runsystem' isn't getting called <bjc>is is /hurd/ext2fs responsible for doing the initial dirty check and spawning fsck? <yelninei>i guess then wed also need to patch the prefix for the format to /run/current-system/profile/sbin instead of /sbin but I am not sure if that exists yet at this point <janneke>i'm pretty sure ext2fs.static does the dirty check <janneke>good question, no idea if it tries to spawn fsck <janneke>yeah ext2fs/hyper.c has: ext2_warning ("FILESYSTEM NOT UNMOUNTED CLEANLY; PLEASE fsck"); <yelninei>when is /etc/fstab created? it should run as part of activation-service but when does that run? <janneke>well debian calls ext2fs.static with --readonly <bjc>should only mount rw when it's clean <Noisytoot>how can I install the "English (GB)" language for librewolf in guix, or at least set the date format to something sensible like yyyy/mm/dd or even dd/mm/yyyy? only "English (US)" is showing up in the list of languages (unlike in icecat, where all the languages are listed) <Noisytoot>I really hate middle-endian mm/dd/yyyy dates <Sven792>I'm also noticing now that LibreWolf isn't using my system language at all unlike IceCat <Sven792>I tested the flatpak release of LibreWolf and the same result. So not a distro issue but I don't know if it's intended? <futurile>there's two sections of the manual about it, depending on how deep you want to get <bjc>yelninei: activation happens in the boot script <bjc>the loading '/gnu/store/358n1pq6xyccih26z4w575ahqazcah3x-system/boot'... <bjc>that gets called from the rc script which is made from âboot-hurd-systemâ <Noisytoot>futurile: yes, librewolf simply does not respect it <untrusem>Noisytoot: So LibreWolf just to randomize it to reduce fingerprinting. You should check the settings. <Noisytoot>untrusem: It's not randomized, it's set to English (US). And I did, the only option in the menu is "English (US)" and "Search for more languages" results in "LibreWolf canât update your languages right now. Check that you are connected to the internet or try again." <Sven792>untrusem: that's not really true. For this TorBrowser and IceCat will ask you explicitly the first time you're using it. The interface will be native though <Noisytoot>There's a 'Use your operating system settings for âEnglish (GB)â to format dates, times, numbers, and measurements.' option which seems to partially fix the issue (I just had to refresh the page I was testing on) but it's still using mm/dd/yyyy dates on "certificate expired" pages. That's probably a different librewolf or firefox bug though <Sven792>I tried to change my system language to asian ones out of curiosity. Japanese (JP) and Korean (KO) gave me build errors and with Thai I only see placeholder letters everywhere (no font installed I guess). ;D <Rutherther>Sven792: right, this should be fixed in next release though, the fonts have been added. Last release, 1.5.0, doesn't have them, though. <Rutherther>although I am afraid the images are broken right now and you can't install from them, so... <Rutherther>as for build errors, what errors were those - are you sure you entered an existing locale? <Sven792>Rutherther: oh that's probably 100% my fault <attila_lendvai>there's no virtiofsd packaged for guix? or how do i use it? should i just stick to 9p for shares between the host and the vm? <bjc>ok, i can confirm that runsystem isn't being called when the fs is dirty. also, just adding â--readonlyâ to the ext2fs command line causes failures from ârun-consoleâ <Sven792>whiteshark: Funny but is this part of the Dead Internet Theory? ;p