<luciddreamzzz>hi is there something i can do to fix: guix substitute failed to download <emacsomancer>rekado talk looks very interesting. I'll be looking forward to watching the video too. <lfam>apteryx: I probably tested it wrong. I used `./pre-inst-env guix build -S ...` but I should have probably `guix pull`-ed from the Git tree <lfam>apteryx: It works with Ludo's suggestion <apteryx>lfam: how did you tested it? it's a world rebuilding change, right? <lfam>apteryx: No, it doesn't require rebuilding. It builds things we know the hash of in advance (fix-output derivations) so we don't have to rebuild what it depends on <lfam>I mean, we don't have to rebuild what depends on it <apteryx>still seems a bit tricky to reproduce; I either get a substitute, or it goes off building openssl with --no-substitutes <apteryx>I'm using this command: ./pre-inst-env guix build -S libssh --no-build-hook --no-substitutes <apteryx>I would have thought it would have tried building only libssh, not its transitive graph. <apteryx>^ it seems it wasn't building the transitive graph (i'm not sure why it was building those two dependencies though) -- later it tried downloading substitutes other than libssh, which was what I would have expected. <apteryx>seems I have to reboot to have a new system's mcron jobs refreshed ***astronavt_ is now known as astronavt
<apteryx>oh, it worked at my last reconfigure (cron jobs are refreshed). nevermind <g_bor_>I am making some progress in making icedtea reproducible. <g_bor_>However there seems to be some files that have interesting differnces. ***roptat_ is now known as roptat
<lfam>apteryx: It's a bit tricky since we patch the libssh source code, so `guix build --source libssh --check` doesn't necessarily trigger the download, but only the patching. You'd have to garbage collect the Git checkout from your store in order to make Guix download it again while testing <lfam>Once you do that, you can do it again with substitutes until it starts the Git clone, stop it, add --no-substitutes, and then it should do the right thing for your tests <dustyweb>I tried doing a guix pull, it started compiling guile-2.2.4, and got stuck on running tests all night <dustyweb>I don't remember guile's tests taking so long. ***jonsger1 is now known as jonsger
<wingo>dustyweb: i think there is a bug in the sandbox test that makes it hang sometimes <joshuaBPMan>morning, I installed guile-charting just now via guix. and (use-modules (charting)) seems to fail. <joshuaBPMan>I've installed guix in a foreign distro. Is there something that I need to set? <roptat>environment variables, guix should have tell you something about it? <roptat>you can use `guix package --search-paths` to learn about what environment variables are required <roptat>maybe you need to install guile through guix too <roptat>I think it can work with your system's guile, but I'm not a fan of mixing guix libraries with system executables in general <roptat>really, you should only have to set GUILE_LOAD_PATH=~/.guix-profile/share/guile/site/2.2 <linschn>I'm trying to start the network with a static IP address (herd start networking-ens3) but I get "In procedure inet-pton: Bad address". <linschn>The network was configured like so in my system declaration: <linschn> #:name-servers '("8.8.8.8" "8.8.4.4")) <linschn>I don't know how to check if this configuration has actually been applied. <ngz>The 291.168 looks suspicious <lfam>It's not a waste of time. We are here to help :) <jlicht>Can anyone recommend a VPS hosting provider that offers GuixSD-based machines? <jlicht>mbakke: I sadly have no spare hardware, or at least none that is actually connected to the web on an unmetered connection :-) <ng0>jlicht: out of the box no one <ng0>you can try to convince hetzner for community ISO inclusion (I didn't want to go further) or hit up 1984hosting, who were pretty open to the idea and we might even be back to the ability to run on their lowest tier <jlicht>ng0: I'll at least put it on my list. I'm not quite sure GuixSD is going to fit on 512MB of RAM though ***jonsger1 is now known as jonsger
<g_bor>anyone with more recent guix pull transition experience? <roptat>hi g_bor I read earlier you are making progress on icedtea reproducibility? <roptat>I just did a guix pull from an old one and it worked perfectly <g_bor>roptat: Maybe you could help me out, there is a thread on the ml: [outreachy] Further steps <g_bor>there was some problem with guix pull. <g_bor>roptat: yes, the doc output is now reproducible, but jdk is not. <g_bor>This is not pushed yet. Waiting for core-updates. <g_bor>I can push somewhere if you would like to have a look. <g_bor>also it is only for icedtea-6 as of now. <roptat>g_bor: but the jdk was reproducible before, what happened? <g_bor>it seems that it was not for icedtea-6. I thought only icedtea-8 jre was reproducible. <g_bor>It also seems that when there is a difference in outputs, then only then first differing one is reported (or was reported, I don't know if this has been addressed) <Laalf>I see lxqt is getting packaged. Does anyone know how far it is? <jlicht>jonsger: that seems very promising, thanks! Their pricing model also seem very fair <g_bor>It contains zip files with timestamps, I am fixing that right now, then there are some more interesting things, I guess I will need some help there. <g_bor>One is ct.sym, I guess I will have to write a utility function to rewrite the timestamps unsing source_date_epoch. <g_bor>Another thing, is there is some stuff in a file where a hex something is after a jamvm error... <g_bor>Had no time yet to investigate that further. <g_bor>On guix-help an indeterministic test-suite failure was detected in qemu-minimal. <efraim>nss is failing for me on core-updates for armhf-linux <g_bor>Laalf: as far as I know, it is effectively complete, it is being reviewed right now. Reviewers appreciated. They have some problem with panel, that is marked as help wanted. <Laalf>g_bor: thanks for the info. Is a service for lxqt specified yet? If so I think I will have a look at it since a friend of mine is interested in getting into guixsd but has no DEs that really suit him <g_bor>Laalf: I don't think so, but I don't know for sure. <Laalf>Alright. I will have a look at services when I am back on my PC. I never looked where they are specified. <linschn>I'm the one who reported the indeterministic test-suite failure for qemu, If you need anything (log files or whatever) hit me up by email or through here <g_bor>linschn: Thanks for assistance. I believe, that unfortunately all we have is the log you sent, so someone will have to reproduce the issue using guix build -K to investigate further, or have a look at the code. <g_bor>From the error message it seems, that there are randomized test in that module, which should be disabled. <roptat>Laalf: it may be that there is no need for a service <Laalf>roptat: for xfce you add xfce-desktop-service to your services so i expected it to behave the same way as xfce, gnome, enlightenment, mate.... why wouldnt it need one? even if its not needed a service for consistency would be good <janneke>any advice on installing GuixSD at this moment in time? I don't have enough information for a proper bug report yet, but there was an attempt to build bzip2 from source and on 0.15.0 the source tarball fails to download <roptat>Laalf: you're right, lxqt is a DE, not a WM <Laalf>roptat: i was searching for a long time for the services that bspwm needs and never found it <roptat>Laalf: in the case of a wm, the login manager is able to find it directly without a service <roptat>that's probably what happened :) <Laalf>roptat: of course they do. i now run bspwm. but i expected wms to need services as well. well no <ng0>some of the login managers - that is, all of the ones we have service files for - pick up .service files which almost in all cases work ***nly is now known as nly_
***nly_ is now known as nly
<Laalf>how do you theme qt applications under guixsd? qt5ct is not packaged. ***nly is now known as pig
***pig is now known as nly
***nly is now known as y
***y is now known as nly
<jayspeer>hi, how is hurd support going? I heard that nix is already on it <vagrantc>wow, that would be odd if nix gets decent hurd support before guix :) <efraim>and of course now nss for armhf builds on core-updates <civodul>vagrantc: turns out i know very well the person who added Hurd support in Nixpkgs ~7 years ago :-) <civodul>i doubt it has been touched in the meantime <civodul>jayspeer: but yeah, Guix allows you to cross-build to GNU/Hurd (from GNU/Linux), and it's also supposed to run natively as well though i suppose it suffers from bitrot <buenouanq>I got an old Dell tower for free and it's been running dist reconfigure for \emph{days} now... <buenouanq>I was getting a bunch of gateway timeout errors or something so I had to use --fallback <civodul>i sometimes feel lonely on guix-patches :-) <civodul>well i'm not saying that to make you feel guilty <civodul>it's more of a call for everyone to join the reviewing party! <davexunit>civodul: one day I'll return to guix... one day... <davexunit>but for now I'm just a user with commit access that occasionally updates something ;) <jonsger>hm. building the kernel after "guix system init config.scm /" on a fresh ubuntu. gonna be expensive, the cloud... <civodul>are you getting substitutes from berlin? <janneke>it seems that installing GuixSD atm is pretty troublesome <janneke>i tried to help a collegue today to do so and i don't have enough info to create a bug report, but there were many hurdles