IRC channel logs
2025-07-27.log
back to list of logs
<ekaitz>Kolev: in my network the substitutes work perfectly fine <ekaitz>have you tried to --no-substitutes? <ekaitz>it might be a network issue in your side <Kolev>ekaitz, isn't --no-substitutes bad?? <Kolev>Firefox: The server at bordeaux.guix.gnu.org is taking too long to respond. <Kolev>ekaitz, well, building everything from source sounds bad. <ekaitz>that's an opinion, not a fact. It has pros and cons really <Kolev>Trying --no-substitutes i guess <User-42>i have a usecase i would really like to make work, i've just installed 1.4.0 on a new machine, and i want to reconfigure it to be able to be a guix deploy target. i do not want to first guix pull, i simply want to reconfigure with ssh and signing keys in place. However i cannot get the guix system reconfigure to stop running 'guix pull'. Is there a <User-42>way to force guix to reconfigure with the current system guix? <Kolev>ekaitz, building stuff from source takes even more time <ekaitz>Kolev: but it might work, even if it takes long... what you are doing now doesn't even work <ekaitz>User-42: doesn't reconfigure use the current guix? <User-42>i thought so too, but it git pulls immediately, even when re-using the /etc/config.scm <Kolev>Well, it's building stuff... <ekaitz>User-42: manual says: It is highly recommended to run guix pull once before you run guix system reconfigure for the first time (see Invoking guix pull). Failing to do that you would see an older version of Guix once reconfigure has completed. <User-42>i mean, i'm not _sure_ it runs guix pull, but it starts "recieving objects" <User-42>if i may ask, what is it doing then? <Kolev>At least --no-substitutes is working... : ( <Kolev>Hey, it says it's building a kernel. This is scary. Kernels take a long time to build. <ekaitz>User-42: guix pull is... literally a git pull behind the scenes, that downloads the new guix code. <User-42>yes, i mean the other way around, this seems to be a git pull, which i assumed meant a guix pull as well <ekaitz>Kolev: relax, it wont hurt. It will just take some ours from your server's time. Leave it there. <User-42>i just don't understand guix well enough to see why a reconfigure would need to "receive objects" for 30m <ekaitz>User-42: it's a git pull, but I don't really know of what. Probably pulls guix, but doesn't change the commit id you are in? idk really <User-42>yeah, i was hoping to avoid git pulling at all to have a smoother install flow, but i guess i'll have to find another way <ekaitz>User-42: i would just reconfigure, it doesn't do that every single time <Kolev>ekaitz, I switched substitutes to a North American mirror. <Kolev>Oh no! I forgot to do `sudo`! <Kolev>Next: How to push local HTML to /var/www/. <RavenJoad>Kolev: I actually have a blog post about using Guix to push a website to /var/www so that nginx can see it and that the website source can be built by Guix, if you're interested. <RavenJoad>Sorry for the self-plug, #guix, but I did a deep-dive a few years back on making nginx work nicely with Guix-built websites. <Kolev>RavenJoad, I don't see 80 open on the final config. <RavenJoad>Modify the `listen` field to have "80" in the list of open ports to listen on? Otherwise, I think certbot will add a redirect server to make anything that requests an HTTP (80) port use HTTPS (443). <Kolev>RavenJoad, I worry that certbot won't be able to renew, without 80 functioning. <RavenJoad>certbot is capable of running their own HTTP server to get/renew certs. There's a --standalone option which "Run a standalone webserver for authentication". <Kolev>RavenJoad, is that what your code does? <sughosha>HI all! How can I identify an input that is an origin (not a package), which is declared like (list (origin ...))? I want to access it with (assoc-ref inputs ...). <sughosha>I remember that there was a commit in the guix repository regarding what I am looking for, but I forgot what it exactly was. <tusharhero>Hello, can someone explain how we can enable plymouth for Guix? It is available as a package. But there is no service for it it seems. <ArneBab>I’m getting build errors at the moment. For example xsane does not find gimp-2.0 -- is that already known or is it just my system? <namewillforget>Hey everyone! I'm a bit confused because my system config is always opened in read-only-mode. That's even after I copied the original file and started to use the copy. I thought I'm supposed to edit that file directly and use that file to configure my system. Why is it in read-only? <sham1>Look at the file permissions <namewillforget> It's owned by root, its group is root and it is read-only: -r--r--r-- <sham1>namewillforget: you want it to be at least -rw-r--r-- <sham1>Then it won't be opened read-only <gabber>ArneBab: mine fails with "SANE-1.0.0 or newer is needed for compiling xsane" <gabber>no, you were right. "No package 'gimp-2.0' found" <ArneBab>cow_2001: I feel your frustration. At work we have the rule that only those linter rules are allowed to be mantadory that have no existing violations in the code base. All the others are warnings to encourage fly-by fixing them, but not required. <zamfofex>cow_2001: I don’t think your justification for the comments makes sense. A comment doesn’t need to be inside anything, its “scope” has no bearings on anything. It just needs to be close to the code it’s explaining. <sham1>Time to build Emacs from sources because there's no lucid variant of emacs-next... great <sham1>Or rather, use (emacs->emacs-next emacs-lucid) to create a new package definition. Hopefully my computer doesn't catch on fire <dariqq>anyone using pam_mail? it used to set $MAIL to /var/mail/$USER but now it is /var/spool/mail/$USER and I am not sure what changed <dariqq>i just changed it back with dir=/var/mail but I would love to know why this is necessary now <sham1>I was under the impression that putting the user mail directory under /var/spool/mail is the standard one <dariqq>maybe something related to core-packages-team merge? my generation from before the merge has it as var/mail but afterwards it is var/spool/mail <dariqq>i think i found it: PAM_PATH_MAILDIR now defaults to /var/spool/mail (the fallback) because it fails to compile the test program to check if _PATH_MAILDIR is #define'd: implicit declaration of function 'exit' [-Wimplicit-function-declaration] <dariqq>uhoh: Building the following 2979 packages would ensure 7239 dependent packages are rebuilt <apteryx>hako: hello! I'm trying to use the new cargo-inputs & tooling to refresh the crates for librsvg 2.60.0. Any reading/doc, or tips? <apteryx>looks like 'guix import' knows to call the right scripts <apteryx>looks like it's not 'guix impor crate', but some new lockfile importer <apteryx>I think the entry point for the new crate lockfile importer is something like this? ./pre-inst-env guix import crate --lockfile=/tmp/Cargo.lock --file-to-insert=gnu/packages/rust-crates.scm librsvg ? <apteryx>after that, how do we add crate origins to (gnu packages rust-sources) ? is there something automatic? <noe>apteryx, lilyp for the cargo+meson apps of GNOME 48, should we wait for rust-team merge as they have all the rust-gtk4 and related dependencies updated? Or do them but on rust-team branch 🤔 <apteryx>I'm attempting to update librsvg to 2.60.0 on my WIP branch <apteryx>i've imported some of the tooling needed from rust-team (cherry-picked) <apteryx>do we need rust beyond librsvg for gnome 48 ? <noe>gnome-logs, gnome-tour, loupe and snapshot all use rust, usually rust-gtk4-0.9 <noe>as well as many gnome circle apps <noe>actually not gnome-logs <apteryx>I see. I guess we can wait for now, or add what's missing or rust-team branch? or rebase or we could rebase gnome-team on their branch and work it there. <apteryx>(on gnome-team), if it's just adding gnome stuff on top of it <apteryx>this seems to have work for getting the needed librsvg crates in: $ ./pre-inst-env guix import --insert gnu/packages/rust-crates.scm crate --lockfile=../librsvg/Cargo.lock librsvg && etc/teams/rust/cleanup-crates.sh <kratacoa>Hello, has anyone here had issue running Picard? It tells me that it cannot find `wayland-egl` among Qt platform plugins. I've run into a similar trying to build `kitty`, with mesa/lib/libEGL.so1 failing at the `setup-py` command <kratacoa>without much thought, I've tried to mishmash packages related to egl, libgvnd, qtwayland, etc. <kratacoa>This is being done on a laptop powered by AMD Ryzen 5 PRO 4650U <kratacoa>in a sway setup, without any DE installed <kratacoa>ok, for Picard I somehow managed to make the run app by erasing the environment variable QT_QPA_PLATFORM=egl-wayland <lilyp>noe, apteryx: given the layout change, I'd wait for gnome-team except on the most important things that are already packaged <sneek>Welcome back lilyp, you have 1 message! <sneek>lilyp, apteryx says: heads-up; I've rebased gnome-team on current master <lilyp>if only one or two additional crates are needed, then fair enough, but we definitely don't want to duplicate our efforts <lilyp>I think getting a head-start on rust-team and starting to package stuff there or cherry-picking some of their commits might also be fair game <noe>I think the second option is easiest (putting the packages on rust-team) <lilyp>apteryx: what's the build status on gnome-team rn? do we at least get our gstreamer things and libadwaita? <umanwizard>Hi all. Anyone know why sometimes when building packages, guix starts randomly emitting stuff from the middle of the build log? Example: https://bpa.st/WBGQ <umanwizard>And always prefaced with strings like: ��@ build-log 7460 1316 <kratacoa>umanwizard: that looks like gcc output as it compiles <umanwizard>kratacoa: that is true, but doesn't answer my question <umanwizard>kratacoa: notice nothing was printed (as expected) for gupnp-dlna, gusb, or ibus. Then randomly, it started printing stuff partway through the build of inkscape. <noe>umanwizard, happens to me too sometimes <umanwizard>(I find that usually "guix does weird stuff" is attributable to me being on aarch64, though that doesn't seem likely to be the cause of this) <jackhill>how should I go about getting openssh to listen on multiple ports? <Kolev>jackhill, would you have to have multiple openssh-service-type instances? <nomike>I have submitted a PR for an update to package OpenSCAD and I'm working on a package PythonSCAD right now, which inherits from OpenSCAD (they're separate projects, but PythonSCAD forked from OpenSCAD so inheriting makes sense). This however, creates a dependency as the new PythonSCAD package depends on some structural changes made to the OpenSCAD one. <nomike>Should I wait until OpenSCAD is merged before submitting PytonSCAD, or could I branch PythonSCAD off of the OpenSCAD branch and sumit it right now already? <Deltafire>thanks for updating it btw, i use openscad quite a bit :)