IRC channel logs
2024-06-29.log
back to list of logs
<JetpackJackson>im gonna copy the checkouts folder from /root/.cache/guix (which has both channels) to ~/.cache/guix <JetpackJackson>drat ok it still doesnt update so i think im gonna check my git config cause the error does say "git error" <ieure>Yeah, maybe root's ~/.ssh/config has stuffy your user doesn't? <vagrantc>hrm. i am trying to inject some randomness into a package definition ... it takes a --with-os-version= configure argument ... <vagrantc>from the guile REPL this *seems* to do what I want ... (string-append "--with-os-version=" (number->string (random 100))) <vagrantc>but that bombs out when injecting into a package definition configure arguments ... <vagrantc>hrm. i thought string-append outputs a string? <vagrantc>is ... (string-append "x" "y") != "xy" ... ? <vagrantc>will have to poke at this some other time ... <JetpackJackson>strange, im getting "(exception misc-error (value #f) (value "no code for module ~S") (value ((packages arduino-cli))) (value #f))" when i sudo guix pull my updated channels file with my packages in it <JetpackJackson>but ive organized it like how others have organized it so idk whats going on <freakingpenguin>JetpackJackson: directory is added to the load-path root, not counting that directory. So you're trying to define a (pita-packs packages arduino-cli) in a GUILE_LOAD_PATH/packages/arduino-cli.scm file <freakingpenguin>You'd either need to remove directory in .guix-channel or remove pita-pack from your define-module declarations. <jfred>Hmm. I've been having Shepherd hang sometimes after "guix system reconfigure"s, and there's issue #67538 which seems similar but also was fixed in 0.10.3... <freakingpenguin>Possibly fixed, #65463 was similar and seems fixed but I never figured out why <jfred>so I wonder if it's one of the previous issues not having been fixed or a new one (or something specific to my setup) <freakingpenguin>It's possible the bug's still there and just went into hiding for a bit. I wonder if the symptoms in 65463 match what you're experiencing (although that's the home shepherd, not system) <jfred>you know now that I think about it I think I've had this happen on my other laptop on a home shepherd instance too <jfred>I was troubleshooting a bunch of other stuff at the time so didn't look into it in too much detail, but maybe... <jfred>guess I've got a few reported issues to look through next time and see if my issue matches any of them <freakingpenguin>Best of luck. These bugs seem to come and go and seemingly nobody's never identified the cause. It'd be lovely to iron them out. <JetpackJackson>yknow, when i git clone/push to my repo, i always have to use my ssh key to do so. maybe guix pull is getting stuck at that part <JetpackJackson>good news: removed the passphrase on the key and got the repo to download <JetpackJackson>bad news: repo doesnt show up in guix describe and i cant install any of the packages from it <freakingpenguin>guix time-machine -C channels.scm -- describe doesn't give the right output? <JetpackJackson>without sudo it does that git error again: "guix time-machine: error: Git error: failed to start SSH session: Unable to exchange encryption keys" <JetpackJackson>i guess tomorrow ill figure out why sudo is the thing that makes stuff work <freakingpenguin>I'm not sure what the problem is but hopefully that helps you narrow it down to "probably not the channels.scm or repo itself". <JetpackJackson>guix pull -C channels.scm does not work so i guess its something else then <JetpackJackson>especially since i can install the pita-packs packages in the root profile using time-machine <adanska>hmm, `xdg-desktop-portal` seems to have a flaky test in the `test-portals` check. for me, it failed on the `/portal/inhibit/delay` portal check, but passed the second time. It seems that its failed on CI and the substitute isnt avaliable. <adanska>oop, now its failing on `tests/test-document-fuse.sh`, `capsh: command not found` <adanska>it seems like it needs a `libcap` in the inputs (or native-inputs, not sure if its needed outside of building/testing) <andrewzhurov>Heyo! Why does `guix gc` deletes packages that are needed by `guix`? So you call `guix something-something` afterwards and it downloads freshly removed packages. My expectation would be for it to keep deps of current roots. Is it intended? A tip on how to fix? <ngz>andrewzhurov: It preserves inputs and propagated inputs, but not native inputs. It may be downloading native inputs required to build something. <ngz>It depends on the `guix something-something` <andrewzhurov>Aha, thanks. I noticed rde rebuilds `chromium` and `node` (it's hard to miss, takes hours:), despite them being in profile. But maybe it builds a new version of them, didn't note, so nevermind. <andrewzhurov>As to `guixc smth-smth`, I ran `guix gc -F 20G` which deleted `meson`, after that running `guix system delete-generations` downloaded it back again.:) <PotentialUser-23>im a complete noob, i want to unset GHC_PACKAGE_PATH, how with env path so its permanently unset in user env? <PotentialUser-23>i think guix makes this read only, so i think, correct if wrong, that it must be done within guix home somehow, but i dont know how. <futurile>PotentialUser-23: you said you're a 'noob', do you understand how shell environment variables work generally? <PotentialUser-23>dont know exactly where they're stored, or how programs access them, or how guix changes that, and so forth. <futurile>PotentialUser-23: why do you want to unset GHC_PACKAGE_PATH? <PotentialUser-23>ย ย home-environment-variables-service-type '(...)), but now im having trouble getting rid of it <PotentialUser-23>cabal: Use of GHC's environment variable GHC_PACKAGE_PATH is incompatible with <PotentialUser-23>Cabal. Use the flag --package-db to specify a package database (it can be used <futurile>well you could (a) remove the variable with the simple service thing <PotentialUser-23>i can run a script that runs unset but then its set again on the next call <thanosapollo>how can I disable substitutes and build everything from source, after already installing guix <thanosapollo>couldn't find it on the docs, only info on foreign distro by restarting the daemon <rrobin>there is a --no-substitutes argument <weary-traveler>guix pull seems to be lagging from current state of guix git repository. is this expected? <weary-traveler>i thought if --no-substitutes is given it should fetch the latest <cbaines>what do you mean by lagging weary-traveler? <weary-traveler>cbaines: git is at b6ef3cfdf0e948f3f5e291cbfefea5e5a13068a7 but guix pull is fetching 78b881722f08325c76096557313a0faee847c66f <cbaines>weary-traveler, do you have this file ~/.config/guix/channels.scm ? If so, what is the contents? <cbaines>weary-traveler, looks like you've configured guix pull to sometimes not fetch the latest commit from the Git repository? <weary-traveler>cbaines: it seems i've misunderstood that snippet then. it's effect isn't undone by --no-substitutes? <cbaines>nope, all that'll happen in that case is guix pull will pick a commit with substitutes, but not use them <weary-traveler>i see. so the only way is to not have that in channels.scm for the time being, iiuc <thanosapollo>but I was just looking to disable substitutes entirely, I think removing substitute urls might do the trick <JetpackJackson>freakingpenguin: ok i think i fixed the guix pull issue: i had a line in my git config that forced codeberg to use ssh instead of https... idk how i missed that lol <rrobin>thanosapollo: guix-daemon has the same option, or if you are using guixsd you can configure the guix-service-type with use-substitutes? #f <thanosapollo>I was going for setting (substitute-urls '()) didn't see the use-substitutes?, you saved some headache, thanks! <freakingpenguin>JetpackJackson: Nice! It's always the one thing you configure and forget about. <faust45>I have install guix-1.4.0 on my laptop thinkpad x13 but couldn't boot the system <faust45>udevd[272]: no sender credentials received, message ignored <faust45>Error: Driver 'pcspkr' is already registered, aborting... <faust45>hci0: fails to request rampatch file: /*DEBLOBBED*/ -2 <faust45>hci0: Missing Free firmware (non-Free firmware loading is disabled) <jonsger>faust45: I guess you need non-free firmware which is not part of guix <JetpackJackson>freakingpenguin: indeed lol. im just happy it wasnt something more involved <faust45>but with nonguix I have the same errors <cbaines>hci0 seems to be bluetooth related, and if it is I doubt it's preventing boot <freakingpenguin>faust45: Google associates "fails to requets rampatch file" error with a bluetooth/wifi issue. You could try booting without the m.2 card inserted although I don't understand why that would halt boot. <faust45>freakingpenguin I "fails to requets rampatch file" can be critical for boot process? <freakingpenguin>faust45: I don't know but it would surprise me. The installer image booted fine, right? <dmirth>Hello ... I have a sort of off-the-wall question: I'm doing a manual guix installation which requires that the HDD partitioning be done manually, obviously. But is it possible to script that somehow within guix? <dmirth>Any suggestions would be welcome <faust45>freakingpenguin: yes boot from image on flash was success <freakingpenguin>dmirth: I don't know how to handle it at present, but in the future with #69090 merged and #71657 is fixed you might be able to just flash an image to the disk that'll resize itself. <fnat>dmirth: One of the first things I usually do is to enable SSH access to that I can then do most of the installation from a remote machine. This doesn't answer your question but perhaps it might be easier to do the disk configuration from another machine? <dmirth>freakingpenguin: Thanks! I see where you're going with that. I think I might be able to work with that now. <dmirth>fnat: I had not thought of that at all! Good idea ... I'll give it a try as well! <faust45>have you any suggestions, how I can fix "Error: Driver 'pcspkr' is already registered, aborting..." <graywolf>Hello :) Where can I find the output (stdout, stderr) of the activation-service-type? <freakingpenguin>graywolf: I think they'd use whatever stdout/stderr is present during boot-system in (gnu build linux-boot) since they're run as part of boot-service-type. I don't think they'd log to a file in that case since that's runs before rootfs is mounted. <freakingpenguin>There's some odd behavior at present that can crop up under certain circumstances when trying to use nss-certs in a guix shell container since SSL_CERT_DIR/FILE aren't set at present. <freakingpenguin>The summary is you also need to add something like curl for it to take effect that registers it as a search path which can cause TLS breakages in other packages such as rust-cargo that don't register said search path, making TLS operations fail even when nss-certs are present. https://paste.debian.net/1321851/ <graywolf>freakingpenguin: and during system reconfigure it would go where? <attila_lendvai>i start a `guix system --debug=100000 vm ...` and it just quits, without any hint of what the error is. <ngz>cbaines: IIUC, bordeaux is still building previous core-updates branch (revision f14d43c1). Would it make sense to stop it to leave more computing power for other tasks? <cbaines>it's probably fine to let it continue for now, higher priority builds will automatically take over <ngz>Yes, but isn't it a waste since we know for sure the same packages are going to be rebuilt? <ieure>Anyone aware of debbugs.gnu.org MTA problems? Sent a patch in this morning, but it couldn't be delivered, "no route to host." <ieure>Guess it could be a Fastmail issue or something in between, but GNU infra seems more likely. <Guest44>Anyone know how I can configure GNU Guix without X with the spice-vdagent-service-type? <madeleine-sydney>what can i do about a bunch of `fork: retry: Resource temporarily unavailable` and `vfork: Resource temporarily unavailable` errors during a build? some way to lower the number of jobs guix runs? <freakingpenguin>madeleine-sydney: You can limit the number of cores used with --cores, e.g. guix build hello --cores=3 <JetpackJackson>whats the best way to manage multiple guix.scm files from repos that dont provide channels for their packages? <JetpackJackson>i have two guix.scm files that rely on their respective git info to work lol <SomniusX>Trying non Guix on a macbook air mid 2013 with one of my friends dotfiles for system home etc. <SomniusX>Still can't figure how to enable wireless broadcom-wl <SomniusX>Anyone can share some light? ๐
๐๐ <cmiller>Anyone know how I can configure GNU Guix without X with the spice-vdagent-service-type? <sergiodj>hi folks, can I have a review of snac2 (bug #71587), please? I know things are down right now, no rush <cmiller>To my question, I can just use SSH. Don't know why I thought that complicated. <ngz>sergiodj: At first glance, the snac2 package style could be updated: use G-exp, remove trailing #t from phase. Use #$output instead of %output. You also need to explain why you're disabling tests and configure phase. I also suggest (nitpicking) to move inputs below arguments. I also think the description could be better. But it may be just me: I'm not a huge fan of lists as description (sometimes, it is difficult to write it otherwise, <dgr>what todo about build of info.dir.drv failed? <dgr>install-info: /gnu/store/b1dc9qrjaq1gdbrjwjg4z8xxqd6c5kp2-wget-1.21.3.24/share/info/wget.info.gz: empty file