IRC channel logs
2024-06-25.log
back to list of logs
<juli>JetpackJackson: ime the Guix committers and reviewers are super nice. no one will judge you for not knowing exactly what to do :) <juli>plus, first-time submissions get passed to the mentors team who are specifically geared towards helping out new contributors <ngz>Ah! "tex-team" objective completed! <JetpackJackson>is there a realname policy for submitting packages? as in, can i submit stuff under my username or do i have to use my real name <mekeor>JetpackJackson: no one will doubt whether the name you are using is real :) <JetpackJackson>im currently waiting for the repo to clone, i think ill go take care of some stuff while i wait lol <JetpackJackson>so where do i put the sway-launcher-desktop.scm file that i made in the repo? <JetpackJackson>and when i run mumi new does that automatically add my file to the patch? <lykso>I'm trying to manually populate a partition via a partition initializer function. Everything appears to be winding up in the right place, except it seems the binaries in the packages reference a different glibc than the one that gets copied into the partition. I'm just using g-exp references to packages imported via "use-module". Anyone know why this might be? <JetpackJackson>i guess ill have to try this tmrw cause i dont want to break things <JetpackJackson>should i put it in packages/xdisorg or should i put it in its own file? <JetpackJackson>drat i get "Unable to initialize SMTP properly. Check config and use --smtp-debug." when i tried to send the patch <JetpackJackson>in the future, whats a good email to use for git send-email specifically? i tried looking it up but couldnt find any info <atugaed>Hi if I type reboot or shutdown shepherd doesn't d anything. <atugaed>With the USB ISO it did so its particularly to my install. are there any necessary services needed to get shepherd to work ? <atugaed>Also with the iso i hâd mouse în the terminal. are there bășic services to get mouse working in the terminal that I'm missing ? <jaft`>atugaed: The commands I usually use are ~loginctl reboot~ and ~loginctl poweroff~; I can't speak to the other commands, unfortunately, but the loginctl ones generally work, for me. <atugaed62>Ty. Does your config make any special mention of extra packages used by shepherd ? <atugaed>Yeah even loginctl reboot doesn't work.it just does nothing. <atugaed>I'm just going to read the manual even herd status returns nothing. <atugaed>Yeah dunno so read around, and reboot and shutdown don't work... Not sure what the issue is... <atugaed>Or maybe reboot is suppose to take 5 years ? <atugaed>Yeah no waiting like 5 minutes after running shutdown and it does nothing. <atugaed>OK so it shutdowns herd, so if I do herd status, I get nothing, but otherwise it does nothing. <atugaed>I know its possible to get this to work (shutdown, reboot) because it works on the iso <atugaed>May have found the answer: elogind handles all that, but I think it needs dbus. <atugaed>Nope, added dbus and still... Shutdown, reboot dont work. <adanska>have the guix sysadmins been notified of the strange state ci.guix.gnu.org is in right now? <adanska>as in the root page cant be accessed, etc <deboh>I read in the manual that if at least one package that is a dependency is updated, then the package is completely recompiled. am I right. if so, why is this so? it must be a lot of compilations and can it be avoided? <janneke>deboh: that's right; the short answer is: yes, it's essential for correctness, it's kind of the essential idea behind the nix thesis that guix implements <PotentialUser-49>rekado: tensorflow from guix-science fails with incompatible flatbuffers minor version. I can not open an issue right now, because my github account has 2fa and mobile is not allowed at work :/ <PotentialUser-36>Hi. How should I use a font in guix when it's not available as a package only as a TTF? On Arch, I'd just put it in /usr/local/share/fonts or something like that. What would you do on guix? <jackdk>has anyone been able to build without substituters recently? I'm getting openssl-1.1.1l failures that seem to be https://issues.guix.gnu.org/56137 . How can I track whether/when this gets fixed? I like that guix takes the bootstrap seriously, so I find the "just use the binary from the substituter" a bit unsatisfying <PotentialUser-36>PotentialUser-36: I mean there must be a "guix" way to do it, right? Have not been able to find something satisfying in the manual <ngz>PotentialUser-36: I think you can extend XDG_DATA_DIRS and put your fonts there. You'll probably need to refresh the font cache, too. <ngz>fontconfig looks for fonts in XDG_DATA_DIRS. <ngz>PotentialUser-36: yw <atugard>Hey if I type shutdown or reboot it just hangs. I can quit that and then herd status no longer does anything indicating that somehow shutdown and reboot just shuts shepherd down but not the computer. Why ? <cbaines>atugard, I think there was a bug in older versions of the shepherd that could cause this <atugard>I'm using newest version, and the shepherd on the iso USB works fine??? <atugard>By the iso USB I mean the most recent stable one. <Deltafire>atugard: i had that issue a few months ago, but I think it resolved itself <atugard>At the user level with SX it just boots into an unusable black screen with no mouse. <Googulator>What's a good way to get a usable "guix shell -D" for packages that override the system type, like ghc@7.0 (which is built as i686 even on x86_64)? <Googulator>Just doing "guix shell -D ghc@7.0" gives me a broken environment containing a 64-bit toolchain <irfus>hi all, how do I go about building emacs-next from the latest git sources? Currently the build fails, and applying the patch from issue#70632 did not help (I might have goofed that though) <ngz>Googulator: I am not sure to understand your question, but wouldn't the --system= argument to `guix shell' work? <Googulator>ngz: that would effectively set up the whole Guix environment for i686-linux, building up everything needed for the active profile, which is a lot more heavyweight than the actual environment Guix uses to build ghc@7.0 <Googulator>ghc@7.0 sets #:system "i686-linux", which is seemingly ignored when you do a guix shell -D <Googulator>Does that mean there's no way to get a shell into the same environment that's actually used to build ghc@7.0? <ngz>"-D" puts you in an environment suitable to build a package. #:system defines how the package is built, which happens as the next step <ngz>Googulator: Can you specify the desired C toolchain with --with-c-toolchain as an option to guix shell? <Googulator>ngz: I wish I knew what toolchain ghc@7.0 uses... <ngz>AFAIU, #:system is used to parameterize %final-inputs. After that point… <Googulator>hmm, seems like it is in fact receiving a correct, 32-bit GCC - but ld, for some reason, is a 64-bit one, and it dies when trying to load GCC's LTO plugin, which is 32-bit <Googulator>Is there a way to specify a package by its store path for guix shell? <ngz>This would be surprising. The store path doesn't quite hold the information required by guix shell, does it? <atugard>Hi I'm trying to get openbox to work for users but it seems impossible. <atugard>Currently I've got a config where I install slim and openbox in the system, but then if I try to log in as my username I just get a black screen. <atugard>I can log in as root and it works fine. <graywolf>Hm, how can I create a configuration with mandatory symbol field? (define-configuration/no-serialization foo (bar symbol "")) (foo) --> #<<foo> bar: %unset-marker% %location: #f> <cbaines>ngz, I imagine that's a data service issue rather than something up with what's being processed, I'll take a look... <ngz>cbaines: OK. Thanks! <redacted>my LD_LIBRARY_PATH environment variable isn't being set. when is guix supposed to set that? <redacted>AFAICT ~/.profile is sourced on login as it should <jackhill>redacted: I'm not sure it's supposed to be set. In what context are you expecting it? <jackhill>as I understand LD_LIBRARY_PATH influences the linker's runtime selection of libraries, but guix binaries should have the full path to what the link against hardcoded. <Googulator>FYI: ghc 7.0.4 as packaged in Guix can build a working ghc 7.2.2, which is building 7.4.2 as we speak <Googulator>7.6.3 should be sufficient for building 7.10.2, which we currently build using a binary version of 7.8 <redacted>jackhill: ah, if it's not supposed to be set, then nevermind <redacted>the binary that would need it was installed by npm and it's also off topic <oljenkin2>Hey, I ran "guix system reconfigure /etc/config.scm" without sudo and got an error at the end. Ran the command again with sudo, but got weird errors. How do i recover? <juli>oljenkin2: that's very weird since you shouldn't have been able to run it at all without sudo <juli>but if you're getting errors before the end, you shouldn't have to recover at all <juli>if you've successfully reconfigured more than once since a known-good state, you can use `guix system switch-generation <generation>' <podiki>note that guix system is atomic: if there were errors than nothing will have changed <podiki>if you run guix system reconfigure without sudo, if I remember it just errors at the end since it can't write the bootloader changes <podiki>if you have errors with sudo, probably nothing was ultimately changed for your system, but we need the actual errors to know. either way, you can always boot to your current (working) configuration from grub <juli>podiki: did guix system previously warn when run without sudo? or have i just remembered to use it reliably enough that i forgot it gets to the end before erroring? XD <podiki>juli: it has been a while, but in the past I know it would do stuff (i'm guessing the "guix system build" part) and error at the end with a permissions error <podiki>you can do lots of guix system without sudo, just the reconfigure (last part of it if I remember) that needs root <podiki>often I'll just do a "guix system build /path/to/config.scm" to do the real work, and later do "sudo guix system reconfigure /path/to/config.scm" which will just write the change and is then super quick <juli>as an ocap gal, i prefer to avoid ambient authority wherever possible, even in trusted code <oljenkins>If I run now "guix pull" I get the error: "guix pull: error: Git error: failed to connect to git.savannah.gnu.org: Connection timed out" <podiki>oljenkins: savannah might be slow/down, I would retry a little later (the web interface isn't loading for me right now either) <podiki>juli: I almost forgot my password when I first switched to guix, since I only ever use it for guix system reconfigure :-P as opposed to other distros whereyou need it for all software management <podiki>it is not just you, I can't git fetch/pull from savannah right now either....hopefully resolves soon <wizard>savannah has been down for at least 4 hrs for me too <wizard>is there any way for me to pull from a backup of the repo? <gjason>FSF is having server issues since 12:30 PM Eastern today, lots of their sites and services are still down. <wizard>darn, hope the sysadmins are staying sane <gnucode>I did start using define-configuration ...it's actually pretty cool.