IRC channel logs

2023-02-06.log

back to list of logs

<stikonas>efraim: git log shows you are the one who did last few rust updates in guix. There are binary blobs in rustc source tarball, do we need to deal with them?
<stikonas>so there are two 4.5 MiB blobs in vendor/vte/vim10m_{match,table}
<stikonas>I just asked upstream to update vte to fix this, and they seem responsive. So I think the new rust versions will not suffer from this
<apteryx>mirai: you want a package that defines the SSL_CERT_DIR/SSL_CERT_FILE search path
<apteryx>that's what OpenSSL uses
<mirai>apteryx: this is the package definition I'm working on https://paste.centos.org/view/96ec8042
<footlong>"Untitled - Pastebin Service" https://paste.centos.org/view/96ec8042
<mirai>how/where should this SSL_CERT_DIR/SSL_CERT_FILE go?
<lechner>nckx / yeah, i know the bot needs a URL blacklist. i was just going to wait and see which ones are needed
<apteryx>can I see just bugs, not patches from issues.guix.gnu.org?
<apteryx>sneek later tell mirai I wonder if you could use the perl build system
<sneek>Okay.
<lechner>apteryx / not sure, but i can see them here https://debbugs.gnu.org/cgi/pkgreport.cgi?package=guix
<footlong>"bugs in package guix -- GNU bug report logs" https://debbugs.gnu.org/cgi/pkgreport.cgi?package=guix
<safinaskar>in what language .drv files are written?
<brendyn>safinaskar, i think its just a simple custom format that came from Nix. the parsers is in derivations.cc
<safinaskar>brendyn: thank you
<Lumine>Good morning #guix
<attila_lendvai>mirai, i was planning to use ddclient, but then i ended up using a simple curl call in a cronjob
<futurile>morning Guix!
<user_oreloznog>hello guix!
<a12l>Can someone please try to reproduce this issue? https://issues.guix.gnu.org/61284
<footlong>"[rust] Segfault when trying to download dependencies using Cargo" https://issues.guix.gnu.org/61284
<a12l>Want to make sure that it isn't just my system
<f3n1x>ahós guixers of the world !
<mfg[m]>How do i use custom channels with the pre-ins-env environment? It seems GUIX_MODULE_PATH and GUILE_LOAD_PATH don't work
<attila_lendvai>mfg[m], i think that should be GUIX_PACKAGE_PATH
<attila_lendvai>even though your naming makes more sense, because i think it's not only for packages, but guile modules in general
<mfg[m]>Ah i didn't know that variable i'll try it :)
<mfg[m]>attila_lendvai: yeah that was the rigth variable, thank you :)
<jlicht>hey guix!
<Lumine>o/
<lechner>hello!
<Lumine>Hello lechner
<jgart[m]>jlicht: hey!
<mroh>Love, Peace & Guix! Good morning!
<sneek>mroh, you have 1 message!
<sneek>mroh, apteryx says: I wonder if you could use the perl build system
<unmatched-paren>hello guix :)
<unmatched-paren>anyone around who could merge my blog post? :)
<mfrancesconi>I am using an ssh account that connects to an external server, i have downloaded through guix some software like samtools and bedtools but when i try to use them in my directory it gives me this error:
<mfrancesconi>-bash: samtools: command not found
<mfrancesconi>In my direcory, however, there is the directry guix.profile and if I go into the bin folder of this, I have everything I downloaded and if I run:
<mfrancesconi>./samtools
<mfrancesconi>In that direcory, I can see it and works. But how I can use that software packages in another folder? I try with:
<mfrancesconi>PATH=$PATH:~/home/hub_galli/mfrancesconi/.guix-profile/bin export PATH
<mfrancesconi>But it doesn't work. What I'm doing wrong?
<drakonis>ah you sent too many messages too fast
<mirai>mfrancesconi: use a paste website
<mfrancesconi>hello, I am using an ssh account that connects to an external server, i have downloaded through guix some software like samtools and bedtools but when i try to use them in my directory it gives me this error:
<mfrancesconi>-bash: samtools: command not found
<mfrancesconi>In my direcory, however, there is the directry guix.profile and if I go into the bin folder of this, I have everything I downloaded and if I run:
<mfrancesconi>./samtools
<mfrancesconi>But in that direcory, I can see it and works. But how I can use that software packages in another folder? I try with:
<mfrancesconi>PATH=$PATH:~/home/hub_galli/mfrancesconi/.guix-profile/bin export PATH
<mfrancesconi>But it doesn't work. What I'm doing wrong?
<lfam>What does `echo $PATH` say? And how about `command -v samtools`?
<mfrancesconi>With echo $PATH I have: /opt/sge/bin:/opt/sge/bin/lx-amd64:/hpc/local/CentOS7/hub_galli/bin:/hpc/local/CentOS7/hub_galli/sbin:/usr/lib64/ccache:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/gnu/profiles/base/bin:/home/hub_galli/mfrancesconi/bin:/home/hub_galli/mfrancesconi/home/hub_galli/mfrancesconi/.guix-profile/bin:/home/hub_galli/mfrancesc
<mfrancesconi>oni/home/hub_galli/mfrancesconi/.guix-profile/bin
<mfrancesconi>and with -v samtools: -bash: -v: command not found
<lfam>Does this directory contain samtools: "/home/hub_galli/mfrancesconi/home/hub_galli/mfrancesconi/.guix-profile/bin"
<lfam>?
<lfam>It looks like a mistake to me
<mfrancesconi>yes, if I go in /home/hub_galli/mfrancesconi/.guix-profile/bin and I run ./samtools it works
<lfam>Do you see the difference in what I wrote and what you pasted?
<lfam>The path has extra "/home/..." sections
<mfrancesconi>oh I see but I don't have that directory
<mfrancesconi>my correct directory is the one that I just post it
<lfam>The problem occurs because you did `PATH=$PATH:~/home/hub_galli/mfrancesconi/.guix-profile/bin`. The "~" expands to "/home/hub_galli/mfrancesconi", so that section is duplicated
<mfrancesconi>Oh, how can I eliminate it? and do it right? or I can just run again the PATH?
<lfam>You can just run it again
<mfrancesconi>Ok thank you!
<mfrancesconi>I export the PATH but the problem doesn't fix it
<lfam>Then, the PATH is probably still incorrect
<lfam>You'll have to double-check that it's correct
<mfrancesconi>if I change the direcorty I can not find my packages
<mfrancesconi>ok
<mfg[m]>Doen anyone know if there is a specific reason why guix shell opencv; ls $GUIX_ENVIRONMENT/include has the structure opencv4/opencv2/*.h?
<mirai>apteryx: setting native search paths to (native-search-paths
<mirai> (list $SSL_CERT_DIR $SSL_CERT_FILE)) did the trick, thanks
<mfg[m]>Ah well, thi seems to be normal it's the same on arch
<mirai>why do some programs end up with paths like these? WARNING: Failed to create cache file /gnu/store/k4nqw1g3qac7g6nj4xw036xw16n6zyp3-ddclient-3.10.0/var/cache/ddclient/.ddclient-real.cache: Read-only file system
<mirai>it tries to use /var/cache but with the store as a path prefix?
<lfam>The program needs to be told where to put the cache
<mirai>lfam: the program accepts a path with the command line
<mirai>but by default it will use /store..../var/cache ?
<lfam>It thinks that / is the store directory
<lfam>It's a common bug
<mirai>is there something patching the simple non-store /var/cache path to /store..../var/cache?
<lfam>It should be fixed in the package definition, at build time
<mirai>lfam: bug? Can you point to examples on what causes this bug or how can this be fixed in package definitions?
<lfam>Try grepping for 'var/cache' in the Guix source code
<mirai>could be worthwhile patching upstream as well
<lfam>I see this package already tries to do something about /var/cache, but obviously it doesn't work
<mirai>oh, I'm updating that package
<mirai>it got support for autotools
<mirai>so I erased (commented that region) the whole arguments
<mirai>yeah, my eyes missed the $cachedir among the sea of red in emacs
<mirai>sneek: later tell apteryx "does perl-build-system work with a autotools project?"
<sneek>Will do.
<mirai>lfam: I see this line https://github.com/ddclient/ddclient/blob/9fccfde9e5f39a91dbce4724fb17ae2e10dcfea7/ddclient.in#L52
<footlong>"ddclient/ddclient.in at 9fccfde9e5f39a91dbce4724fb17ae2e10dcfea7 · ddclient/ddclient · GitHub" https://github.com/ddclient/ddclient/blob/9fccfde9e5f39a91dbce4724fb17ae2e10dcfea7/ddclient.in#L52
<mirai>is this a upstream bug or is this something to be patched from guix side
<lfam>I'd say it's an upstream bug
<lfam>But, we should fix it
<mirai>yeah, I see that localstatedir <https://www.gnu.org/prep/standards/html_node/Directory-Variables.html> says:
<footlong>"Directory Variables (GNU Coding Standards)" https://www.gnu.org/prep/standards/html_node/Directory-Variables.html
<mirai>> $(localstatedir) should normally be /usr/local/var, but write it as $(prefix)/var.
<mfg[m]>Does someone have experience with `cmake-build-system` and knows how to use googletest with it? I get a bunch of errors like: ""
<mfg[m]>Target xy links to target "GTest::Main" but the target was not found.
<Guest46>How can I configure virt-manager with ovmf and tpm2-tss?
<mirai>sneek: later tell lfam I'm not so sure if this is a bug anymore as <https://www.gnu.org/software/make/manual/html_node/Directory-Variables.html> says that sysconfdir and localstatedir are usually under @prefix@ but the description is as if they're supposed to be under /
<sneek>Okay.
<footlong>"Directory Variables (GNU make)" https://www.gnu.org/software/make/manual/html_node/Directory-Variables.html
<GNUmer>Hey! For some reason my Guix install decides to not include doas into /run/setuid-programs after a reboot despite it being added there right after reconfiguring. Opendoas is clearly declared as a setuid program in config.scm but still ignored after a reboot. Does anyone have information on this?
<bjc>i add it to the ‘setuid-programs’ field: http://ix.io/4niR
<cwebber>lechner: it was mainly "this is easy to write, easy to canonicalize, easy for me to understand"
<cwebber>"easy to implement in any language I need to in a half hour or so
<GNUmer>Interesting. Still doesn't work, even though I changed append list to cons.
<bjc>it may require a reboot? i'm not sure; i set this up a long time ago.
<GNUmer>Rebooted it and it hasn't added doas to /run/setuid-programs after the reconfiguration or the reboot.
<bjc>weird. that's literally all i'm doing. the only other refs to ‘doas’ in my config are installing the package itself, and creating ‘/etc/doas.conf’
<GNUmer>Have you installed the package as a system-wide package through config.scm?
<bjc>yep
<GNUmer>Alright, seems weird then since I've declared it as a system-wide package and I've also declared it as a setuid program like the example shows
<lechner>cwebber / thanks! i like the format and may use it. has there been an effort to upload the Haskell package to Hackage?
<cwebber>oh uh... I dunno!
<cwebber>I don't use hackage :)
<sughosha>Hi, is it possible to add the source of a different package as input?
<GNUmer>It's really weird. I can use doas to reboot the computer but it's gone from /run/setuid-programs after reboot.
<bjc>that sounds like the on-boot activation isn't being run for it. i'm really not sure why that would be, though
<mroh>sughosha: see "guix edit vulkan-loader" for an example. It uses the source of googletest as a native-input.
<sughosha>mroh: Thanks! Got it.
<lechner>cwebber / i have guile on one side and, currently, Haskell on the other
<mfg[m]>mroh: incidentally this is also what i need :D (especially for googletest) thanks for the example :D
<mroh>you're welcome! glad I could help.
<mfg[m]>mroh: do you also happen to have an example which shows how to add an input inline? I have a tar archive containing data needed for tests; this should go into native-inputs and be unpacked in the source directory
<mfg[m]>or do i use a snippet for that?
<cwebber>lechner: neat.
<a12l>When I try to run guix shell -f guix.scm I get the error: https://paste.debian.net/1269850/
<footlong>"debian Pastezone" https://paste.debian.net/1269850
<a12l>guix.scm: https://paste.debian.net/1269849/
<footlong>"debian Pastezone" https://paste.debian.net/1269849
<a12l>What I'm doing wrong?
<lechner>a12l / i think your guix.scm has an error in the procedure mve-devel
<barocio[m]>I'm trying to make a guix package for hyprland, but I need to update some git submodules. How can I do that from a guix package definition?
<a12l>lechner: That narrows it down a bit, but I'm having a hard time to interpret the backtrace. E.g. is there any indication about what line that cause the error?
<lechner>a12l / i am not sure if the parentheses around mve-devel (after define) matter, but i would remove them. otherwise, i have no experience with the cargo-build-system. i also have no experience with package definitions that use local files, as opposed to a Git repo with a commit reference
<a12l>lechner: Thanks! Got another error after removing the parenthesizes, fixed that and now it download the deps
<ArneBab>How to compile firefox under guix? I know how to get the sources, but compilation fails with libgcc_s.so.1: cannot open shared object file.
<lechner>a12l / yay!
<lechner>ArneBab / that sounds like you need gcc-toolchain somewhere
<ArneBab>lechner: I’m already getting that, but the build tools do not recognize all the libs.
<ArneBab>lechner: or do I need a non-standard output (gcc-toolchain:static?)
<lechner>ArneBab / i'm sorry. i don't know
<ArneBab>lechner: thank you for trying
<GNUmer>I'm a bit dumbfounded by this doas setuid problem. Should I try to edit the setuid-program variable or submit a bug report? It's getting annoying to reboot my machine multiple times only to not find doas in /var/setuid-programs
<a12l>Do you usually use the language specific build system when developing on Guix, and write a package definition when you want to include the application inside a guix channel?
<a12l>E.g. when I'm writing Rust applications I should use Cargo while developing, and PyPI for Python etc?
<lechner>a12l / probably that's easiest
<lechner>GNUmer / i have setuid programs, but on my system /var/setuid-programs does not exist
<GNUmer>I mean /run/setuid-programs. Sorry
<a12l>lechner: Thanks!
<a12l>Then I know that this https://issues.guix.gnu.org/61284 is actually a bug
<footlong>"[rust] Segfault when trying to download dependencies using Cargo" https://issues.guix.gnu.org/61284
<a12l>is there some way to ping people responsible for the rust-part of guix?
<lechner>pinging is possible, but i have no way of identifying who is (or was) responsible
<a12l>Or does Guix even have such a concept as "responsible for language ecosystem"?
<lechner>we have teams
<lechner>GNUmer / not sure what's happening to you
<GNUmer>lechner: Essentially the setuid-programs declaration isn't persisting after rebooting.
<lechner>GNUmer / but it's there after you reconfigure?
<GNUmer>Yeah, it's there after I run reconfigure but not after rebooting
<lechner>did you share your config?
<mfg[m]>Is it possible to reference inputs with gexps? or do i need to use (assoc-ref %build-inputs ...)?
<lechner>mfg[m] / i think you want #$(this-package-input "guile-lib") or similar
<mfg[m]>I'll try that, thx lechner
<VesselWave>Hello! I try to use seatd instead of elogind, after guix system reconfigure cannot launch sway. Because seatd doesn't create XDG_RUNTIME_DIR. Could someone help?
<lechner>VesselWave / I am not sure if that's a solved problem. Do you use Guix Home?
<GNUmer>lechner: This is barely readable, but this is my config: https://gnumer.cc/guix-config.html
<footlong>"My Guix config" https://gnumer.cc/guix-config.html
<mirai>mfg[m]: assoc-ref
<lechner>GNUmer / that looks more or less right to me, but i have never used doas https://codeberg.org/lechner/system-config/commit/9a5c43e3ec8e558b9156e2825014e7e6270b25ea
<footlong>"Drop setuid for opensmtpd executables; fixed in opensmtpd-service. · 9a5c43e3ec - system-config - Codeberg.org" https://codeberg.org/lechner/system-config/commit/9a5c43e3ec8e558b9156e2825014e7e6270b25ea
<GNUmer>lechner: Weird, still doesn't persist even after setting the same kinds of options as your former config did.
<lechner>GNUmer / some folks mentioned doas here before, but it's not a super popular package, i don't think
<lechner>GNUmer / maybe it conflicts with other settings on reboot, such as sudo, although I wouldn't know how or why
<bdju>are there any plans for macOS support and/or aarch64-darwin (apple silicon) support? as a foreign distro
<mfg[m]>bdju: does guile support these targets?
<dthompson>mfg[m]: yes guile runs on intel and arm macs.
<dthompson>bdju: I can't speak in any specific details but aside from having the developers + hardware to do the porting I think there is or was a toolchain issue where it was not possible to build for macos without some proprietary software.
<dthompson>and I don't know what additional complications apple's strictness around signed binaries would cause. homebrew gets by, somehow.
<VesselWave>lechner: I don't use Guix Home, but if you have some config that can help, I'll try to apply it
<Kolev>Thinking of using Guix just to manage Emacs.
<mfg[m]>dthompson: amazing, didn't know guile supports these platforms :)
<bdju>thanks for the response, dthompson
<bdju>I was using nix on a new macbook and thinking about how much I missed guix and its much simpler commands
<Kolev>bdju: can you port guix to macos?
<bdju>no, I don't even make/update guix packages myself generally... sorry
<dthompson>for guile, specifically, there's guile-homebrew https://github.com/aconchillo/homebrew-guile
<footlong>"GitHub - aconchillo/homebrew-guile: A Homebrew Tap for Guile libraries and applications." https://github.com/aconchillo/homebrew-guile
<Kolev>If guix could run cross platform it would be great
<a12l>Where can I find teams.scm, as described in https://guix.gnu.org/manual/en/html_node/Teams.html?
<footlong>"Teams (GNU Guix Reference Manual)" https://guix.gnu.org/manual/en/html_node/Teams.html
<jonsger>a12l: https://git.savannah.gnu.org/cgit/guix.git/tree/etc/teams.scm.in
<footlong>"teams.scm.in\etc - guix.git - GNU Guix and GNU Guix System" https://git.savannah.gnu.org/cgit/guix.git/tree/etc/teams.scm.in
<cel7t>Hi, I have installed Guix System on a 64 GB partition and there's 11 GB free, however when I'm trying to install or remove stuff I'm getting a symlink error saying the filesystem is full
<cel7t>Any idea how I could fix it? According to the mailing list, it could be because I have exhausted inodes
<lechner>cel7t / which type of filesystem are you using, please?
<gabber>sooo, i've created a patch-series to upgrade Sway (which in turns needed to bump wayland, wayland-protocols-next, wlroots and a bunch of other packages). just to make sure that nothing got broken through the changes i started a bash-script that builds all the packages listed with `guix refresh -l wayland` - i thought that might be a sensible start. so my machine has been building packages (mostly successfully) for a couple of days
<gabber>now. my question: does this make sense? will my machine eventually stop building? should i send in the series or wait for all the jobs to build?
<cel7t><lechner> "cel7t / which type of filesystem..." <- ext4
<lechner>cel7t / what's the output of 'df -i' please?
<KarlJoad>I am working on something right now, where having something like the store monad would be nice. I am performing small transforms on an alist and want that implicitly passed to these functions. Would implementing that like the store monad make any sense?
<KarlJoad>I ask because I may be able to get an idea for what I need to do from Guix's monad definitions.
<jonsger>ACTION pushed jgart[m] lilypond update patch :)
<lispmacs[work]>hi, I'm trying to figure some way to be automatically informed visually when a command-line script finishes running. I use Guix + Gnome DE. Any ideas? Internet mentioned something called "notify-send" but it doesn't seem to be available in my Gnome installation
<pkill9>it's part of libnotify
<jgart[m]>jonsger: awesome! Let me know if you need an extra to review
<jpoiret>gabber: the way to do it would be to send it to the MLs prefixed with [PATCH core-updates], and someone will look at it
<jpoiret>upgrading wayland-protocols is costly
<jpoiret>in the future the process might change, as core-updates cycles are way too long and they're also very hard to merge for the people responsible for it
<gabber>will do!
<gabber>jpoiret: you write MLs - so not just to guix-patches@gnu.org?
<lispmacs[work]>pkill9: thanks