<mala>so I'm doing some minor hacking on a package definition to upgrade it to a latest version, and one of the patches is no longer applying cleanly. My plan is to just download the source and manually create a new patch, stick it into the guix tree and see if it works when I next build it. But is there a better way? Like maybe obtaining or preserving the source tree that guix build sees?
<iskarian>mala, you can do `guix build <my-package> --keep-failed'
<iskarian>it will then keep the build directory in /tmp/guix-build-whatever and you can inspect it
<iskarian>(if patches are failing, it will be the build directory for the source)
***califax- is now known as califax
<zacchae[m]>when booting from a btrfs subvolume, guix automatically adds the necessary subvolume to paths given to the bootloader. Should it also do this for other menu-entries? Because it does
<zacchae[m]>This is problematic when I try to add an entry on a different subvolume
<zacchae[m]>for example, I have guix on subvolume /guix and another OS on /other. I try to specify /other/boot/vmlinuz-linux, but that gets changed to /guix/other/boot/vmlinuz-linux
<zacchae[m]>I'm going to try specifying ../other/boot/vmlinuz-linux, but that seems like a hack mor than the "right" way, no?
<zacchae[m]>Shouldn't the subvolume prefix for guix not be automatically added to auxhilary menu entries? Wanted to ask here before bothering the mailing list
<RRRRedEye[m]>`/guix/other/boot/vmlinuz-linux`, is the boot volume mount at /guix/other/boot ?
*apteryx wonders if a package could be *defined* as cross-compiled
<apteryx>rekado_: there's no need anymore for texlive-union on core-updates; if you remember, the default configuration of texlive was modified so that it could honor multiple TEXMF trees.
<apteryx>the only useful bit of a union that was kept is the generation of font maps; hence it's explicitly named as such now as texlive-updmap.cfg. The commit message of 04a0b1e09abce99857e7930336421ca6d15ae630 has a bit more details.
<zacchae[m]>RRRRedEye: /guix/other/* doesn't exist. I want to specify /other/boot, but /guix is automatically added
<attila_lendvai>i have an app that generates shell scripts (idris). changing it to generate `#!/usr/bin/sh env` doesn't seems to be enough, even though if i look around in the builder env /usr/bin/env is there and seems to work. what am i missing?
<nckx>civodul: Kick them, ban them if their client's too insistent, explain what happened over e-mail & how they can easily get 'unbanned' (and make a note to unban them in a week in case it's fixed) -> is what I do. I don't know who ix is though.
<attila_lendvai>but then how do i solve it? i'll need to pass the shell as a command line argument, or as an env variable? and add support for it to the app?
<rekado_>I also can’t get a substitute for gnome-shell (”server is somewhat slow“) and it tries to build it locally, but fails.
<attila_lendvai>cbaines, these scripts are generated and then run. one backend of idris is using chez scheme, and invokes it through a generated shell script. since then i realized there's already a variable: $SHELL
*attila_lendvai gets into hacking idris to getenv SHELL
<cbaines>rekado_, looking at the code, that error suggests that the data in the cache file is unexpected
<pkill9>please can someone put a link to the guix git repository on the front page of the guix website
<yoctocell>attila_lendvai: You might want to try it to see if things work for you
<attila_lendvai> yoctocell, thanks for the head's up! on a parallel line i'm also working on the bootstrapping of idris itself, i.e. bootstrap it all the way from GHC. but i'll certainly look at them, thanks!
<tschilptschilp23>Hi all! I'm having troubles getting a proper response from python-pamela -- whether I call it from a globally installed one, or one from a venv, I receive a 'undefined symbol: pam_strerror'. If I run pamtester as non-root, it also would not work, as root it does it. Is this a privilege-thing, or am I missing something else?
<pkill9>guile-bytestructures reports invalid target for aarach-64
<rekado_>civodul: is this really a bug, though? When we have a poor wireless connection, is it not possible for these files to be filled with some garbage or at least be truncated and thus unparseable?
<attila_lendvai>yoctocell, most probably only later, but i'm making notes, thanks!
<attila_lendvai>yoctocell, i'm having trouble even with idris 1.3.3 on guix: guix environment --ad-hoc idris clang gmp glibc, then set IDRIS_CC, but inside idris :exec fails with: /tmp/idris28135-0: error while loading shared libraries: libgmp.so.10: cannot open shared object file: No such file or directory. any hints? or pointers to examples?
<civodul>rekado_: no, it's not possible to fill these files with garbage: we only fill them with an sexp written with 'write'
<civodul>so the content is always readable with 'read'
<roptat>it started with the latest version that didn't need a kotlin compiler, and then building the most recent version that compiler is able to build, etc
<roptat>and that's roughly one new bootstrap every two weeks, but I'm hoping it gets better nowadays
<lilyp>Hopefully it's not one new week every two bootstraps like in Scala.
<roptat>scala is worse, it's eventually grounded in non-free software...
<roptat>probably, the best plan for scala is to write an alternative compiler/interpreter in a subset of Scala that can build the full Scala compiler, and create a small Scala compiler in another language
<roptat>that's what we did for OCaml: an interpreter in a subset of OCaml, and a compiler in Guile for that subset
<civodul>and emacs-guix talks to an older 'guix' package, with the previous ABI
<txgvnn>ya, I think so. But I don't know to fix :P
<rekado_>this sounds like the same kind of problem we had to work around with GWL
<civodul>hmm but doesn't emacs-guix run "guix repl"?
<makx>mhhm something seems broken with ci.guix.gnu.org (at least for aarch64); all the recent kernel builds fail
<civodul>i suppose emacs-guix launches "guile -L ..." instead of "guix repl", which would explain why it's a problem
<makx>and if there's something I'd rather not build on a small SoC then itś a kernel
<tschilptschilp23>txgvnn: yes -- that's the one I see on my laptop at home after pulling some two days ago. Here on a fresh VM at work I can't even TAB-complete the guix-*-commands within emacs (tried both putting emacs-guix into config.scm and just guix install as user)
<civodul>makx: hi! do you have a link to a failing build?
<civodul>txgvnn: lemme try updating the 'guix' package; that should fix the issue
<txgvnn>civodul: I see both *Guix Internal REPL* and *Guix REPL* buffer.
<pkill9>how do i diagnose a failure to install u-boot when building a disk- image?
<cybersyn>reza i did it for my first time just using the BTRFs manpages during setup and it was a very smooth and simple process
<tschilptschilp23>txgvnn: Ha - I had to source my .guix-profile, now I'm seeing the guix-*-commands from emacs. I'm getting the error you reported on both of my machines now!
<cybersyn>hi all, i've been off irc for the last few months due to moving into a friends place who's ip was placed on a spammer blacklist by liberachat, but i used to be in here quite a bit and attended fossdem and guix days this past year
<cybersyn>guix is still my favorite thing in the universe :)
<makx>isntall qemu, get mariadb. and postgres. one day i will understand computers
<cybersyn>but i'm wondering, I found out about guix from Brett Gillio who i became quite good pen pals with after meeting him in the OCaml irc. after the emacs conf last year, his blog stopped updating, his git server stopped updating, and he stopped responding to emails. I've been a bit worried that something may have happened to him, and I find myself bumping into his website occasionally.
<cybersyn>i recently saw this DevOpsWorld advert for a presenter that has been doing CI/CD with jenkins and guix. does anyone know where to find the repo? (also, congrats! seems like it could possibly help dispell the myth that guix is experimental software)
<nckx>cbaines: Just FYI your GDS mail was a bit big (1.5MB, soft limit is 1MB).
<cbaines>ah, right, I should have probably compressed the attachment
<nckx>No bigsie. I approved it; wasn't worth a resend.
<nckx>I'm attempting packagement of AspectC++, a Jenga ouroboros that bootstraps itself with a statically linked version of one of its own subprojects, which then makes the whole build itself static too… I think. My understanding of it is currently a hot mess.
<sneek>iskarian, muradm says: also compiled arcan-wayland sucessfully, thanks
<abhiseck>Hello, I want to setup tor daemon and use torsocks... According to some mailing list dicussion, it seems I need to add (tor-service) in services declaration. But which service module provides it?
<roptat>we used to have procedures like (tor-service) that were simply doing (service tor-service-type)
<Noisytoot>I have (service tor-service-type (tor-configuration (control-socket? #t)))
<roptat>everytime you see *-service, it's a procedure you can use, and everytime you see *-service-type, it's not a procedure, and you need to use it with (service *-service-type [configuration]) (configuration is optional)
<abhiseck>Noisytoot: I want specific programs to to work with torsocks, will it do that?
<old>Hey! I have a problem when building a package using the cmake build system. The build failed to download dependencies from github. I get an error message saying `can't resolve hostname github.com. Any idea?
<maximed>old: Add the dependency to 'inputs' (or 'native-inputs' if appropriate)
<sneek>maximed, lispmacs[work] says: I am getting an error when I try to use backquote syntax you defined: In procedure syntax-wrap: Wrong type argument in position 1 (expecting syntax object). Maybe show me how you use it in a definition?
<maximed>old: By design, the build environment does not have access to network
<maximed>Granting access to the network would open a can of irreproducibility worms ...
<old>maximed that's what I though. I tried adding the dependencies to propagated/native-inputs without success. Cmake is still trying to pull de deps from github
<old>I should precise it's trying to pull .tar.gz, and not clone the repo
<maximed>old: Possibly you'll have to patch the CMakeLists.txt to stop it from fetching extra code from github
<Noisytoot>https://web.archive.org/web/19980704181204/http://www.muraroa.demon.co.uk/ says "I'm no longer distributing 0.21, because doing so perpetuates problems with patents, which ensures that the program will never be widely used. That's a shame, because it's a useful program, and lots of people seem to like it. If you use 0.21 already, please upgrade to bzip2. I can't, unfortunately, make bzip2 be able to decompress 0.21's .bz files, since that
<Noisytoot>would render the patent-avoidance exercise pointless. I know changing file formats is painful; from now on, I'll try and make any further changes in a backwards compatible way. ", but I think the patents would have expired by now