IRC channel logs
2025-12-10.log
back to list of logs
<csantosb>sneek, later ask Rutherther, so finally non fixes commits with just upgrades must go to next-master instead of master ? <brendyyn>Has there been any discussion as to the next default llvm/clang version? it's been 4 years since it was last updated and many packages manually use more recent versions arbitrarily. <snamellit>I would like to use SLIB with guile, but apparently the packages does not put it on the GUILE_LOAD_PATH. What is the easiest way to find where in the /gnu/store labyrinth the SLIB library is hidden or transform the package in a manifest to export the GUILE_LOAD_PATH changes? <identity>snamellit: do you have guile in the same profile as slib? <identity>the guile package is the one putting the stuff in the load path, not the slib package; it would be a nigthmare <snamellit>identity: yes, i have guile package. I do not see the slib content in the guile lib directory in the profile. <identity>but do you have the guile package in the same profile? <snamellit>identity: yes, in the same profile, loaded from the same home-profile.scm or guix.scm <snamellit>I distinctly remember using it a while back and I can't remember having to jump hoops at the time. <identity>slib goes into /lib/slib, not /share/guile/site/3.0, so it does not gets put in the GUILE_LOAD_PATH <identity>no idea how that would have worked before <identity>you can add profile/lib/slib to GUILE_LOAD_PATH to make it work, at least <snamellit>yep, found it. I expected it to be in .../guile/3.0 . There are so many places to check. Ok, I can add LIBRARY_PATH to the GUILE_LOAD_PATH. <Rutherther>oh crap did I just break master? I pushed my first commit and I can't pull/time-machine to it, despite guix git authenticate succeeding fine <sneek>Rutherther, you have 1 message! <sneek>Rutherther, csantosb says: so finally non fixes commits with just upgrades must go to next-master instead of master ? <Rutherther>I already tried few commits pushed to my fork and it worked fine, so I didn't presume there could be any problems <cbaines>I get an authentication error at least <andreas-e>Rutherther: Maybe it did not make it? I see the cppcheck commit as the last one on master. <Rutherther>tbh I have no idea what's wrong here, my key is in .guix-authorization, on keyrings branch and it matches the one it's saying it doesn't have <andreas-e>I see the signature. Maybe the keyrings branch is not up to date? Mine is not. <andreas-e>But I get an authentication error also when updating the keyring branch: guix git: Fehler: commit b6472ce4eddb56d80def37340f4a7f35bcbdfdd1 is not a descendant of introductory commit 9edb3f66fd807b096b48283debdcddccfea34bad <cbaines>looks like the commit has been signed with a subkey, and I'm not sure if that's supported <cbaines>or maybe it does work, but you need to specify the id for the subkey used in .guix-authorizations <Rutherther>Anyway can someone get rid of the commit please? I dont think I can force push <cbaines>maybe the issue could be the file on the keyring branch is suffixed .gpg rather than .key <cbaines>I'll give force pushing to master a go, one second... <Rutherther>Thank you very much. Yes I have overlooked that, thats definitely a difference from my own experiments <cbaines>now I'm stuck waiting for Guile to compile a bunch of stuff for some reason :/ <andreas-e>Apart from that, what is different with your key compared to others is that you have several subkeys for signing. But with their different fingerprints this should not pose problems. <cbaines>and my force push failed... I'll need to change the Codeberg config, one second... <cbaines>OK, I think I've managed to force push now, and guix pull is no longer complaining for me <Rutherther>andreas-e: ah you are right. But only one of them should be valid now, rest are expired. I forgot to backup when moving them to a hw key, so I couldnt move them to second hw key afterwards <Rutherther>I was able to successfully get auth working recently with the subkey, so I really dont think it should be part of the issue <andreas-e>Rutherther: I just mentioned it as a result of my looking for differences; I suppose that cbaines's observation is the good one. <andreas-e>Rutherther, I suppose I should push (and thus sign) this commit on your behalf? We can keep the "Signed-off-by" you line, and maybe add one for me. <Rutherther>andreas-e: I would appreciate that, on the other hand it can definitely wait few more hours / days, so it is fine without that <andreas-e>Done! Easier than reopening the pull request... And independent of resolving the authentication worries with your key. <andreas-e>The error message is compatible with cbaines's hypothesis on the wrong key suffix. <graywolf>Hello :) I have a question regarding guix extensions (they do not seems to be documented currently). From what I remember from IRC, there is GUIX_EXTENSIONS_PATH environment variable, that I can set. But I am not sure how to actually write an extension. Are there any requirements on the behavior or interface? I know that for git I can just put `git-foobar' on path and get `git foobar' subcommand <graywolf>working. Is it the same for guix extensions? <graywolf>If there is any documentation I have overlooked, links would be appreciated. <identity>graywolf: presumably you can look at whatever gwl is doing for their extension <graywolf>Hm, I was hoping to be able to drob a binary somewhere, but it seems the way is to create a package and install it. Well, I can live with that I guess. <graywolf>Thanks for the suggestion, will mirror how they do it. <untrusem>has anybody used virtiofsd in virt-manager in guix <untrusem>I have one folder I want to share to my vm which doesn't have internet connection <untrusem>I tried using virtiofs thing but guix doesn't have it I suppose <janneke>when rebasing using magit, i need to: M-x smerge-mode (turn-off), M-x font-lock-mode (enable; why was it disabled??), M-x smerge-mode (re-anable, now with font-lock) -- does that ring any bells? <nbalzarotti>Hi! I've just received my new laptop and wanted to install guix system on it. The installer is currently failing at network connection saying that it cannot connect to the substitute server. Is https://ci.guix.gnu.org down? <meatoid>nbalzarotti: I'm getting a 504 from my work computer here <nbalzarotti>meatoid: thanks, I guess that's the problem. I'll try manual installation with bordeaux only <Rutherther>nbalzarotti mind you that the / returns 504 does not mean the substitute is down <Rutherther>Secondly the check is wrong, it checks the cuirass frontend instead of substitute. You can create /tmp/installer-assume-online to skip this check <nbalzarotti>Rutherther: ok great to know, thanks I'll try touching that file <a4censord>heya, i wanna run a service with shepherd (in user mode), but the thing forks and doesnt have an option no to <a4censord>is there something simpler than writing a script or such and using that to keep track of it? <a4censord>*the thing also doenst write a pid file either <a4censord>i mean the thing uses dbus activation, so i can query dbus if its still running iirk <a4censord>a hmm, the systemd service just does `Type=Oneshot`, so marking it as a run once at start might work <jlicht>Rutherther: do you think the unclean unmount fix(/workaround?) will still make it into the release? <jlicht>this time actually clear: will that shepherd fix still make it into the guix release. <Rutherther>We will base the release off of master except if something went wrong. And still there are pending issues, so still all in master now should make it <Rutherther>There will be a branch off at one point to get no commits at all, but it hasnt happened yet <jlicht>oh the shepherd fix is already in (current master) guix? <jlicht>ah yeah now I see it. Great, thanks! <Rutherther>Yes, for almost a week. If you still experience please report it asap ideally also with your config <jlicht>Didn't reboot yet after reconfiguring. I'll do so tonight (without fear, thanks again) <Rutherther>Note that the first reboot you can stul experience it, because new shepherd is used only after booting <identity>…so reboot with the usual amounts of fear <vntsuyo>are there any GTK+Rust packages available on Guix? <vntsuyo>there are only gnome-authenticator and helvum <jlicht>identity: good quote for MOTD :-) <basicnpc>Is it possible to get a guix shell without my current home dir? It's a mess and stateful. <basicnpc>Thanks! I was looking for the keyword 'home'. <gabber>basicnpc: this confused me too, recently <gabber>maybe we should add change the help text of the -C, --container option to "run command within an isolated container with current working directory exposed" to hint towards this default behavior? <basicnpc>I just tried no-cwd in another dir, but it still shadows the whole HOME. That's indeed weird. <vhns>Hi. I have defined a custom install image a seen here: https://paste.debian.net/hidden/a72bdc47/ . Is there a way to include the derivations/provenance/files for the OS configuration I want to instatiate from the installation media? Such that it won't need to retrieve that much data from substitutes. <vhns>i.e including the stuff from /gnu/store that is a result of my %laptop-desktop definition. <ieure>vhns, I *think* you can include the packages you think you'll want to install in the `packages' field of the operating-system, but I haven't tested this. <ieure>The derivation is computed on-machine, I don't think it needs network other than to fetch the things it's deriving from, which packages should substantially cover. No idea about provenance, but I don't believe that's downloaded either. <vhns>I asked one of the big slop generators to suggest something and it came up with using (computed-file), but I'm not sure how right that'd be. <ieure>I do not recommend asking the machine that pollutes as it lies to you anything, ever. <vhns>whatever, I'll just suffer through waiting for packages to be retrieved during install for now <vhns>19:51 < ieure> computed-file of what? # of %laptop-desktop <ieure>vhns, %laptop-desktop is your operating-system? <ieure>Yeah, that is not going to work at all. Putting packages into the installation image's operating-system should leave them in the store, which should prevent them getting downloaded. I would do that. <ieure>It's at least an approach which works enough to try. Vague suggestions of using computed-file on a thing which is not a file nor needs computation will not remotely work at all.