IRC channel logs


back to list of logs

<zamfofex>Oh, I see a new ‘hurd-packages’ specification on CI!
<nckx>Kolev: Why is your swaylock setuid?
<Kolev>nckx, by default, Swaylock doesn't work. It locks the screen and then doesn't let you unlock. I heard you have to make it setuid.
<nckx>The manual mentions this, so best read the bit about swaylock + setuid.
<nckx>Kolev: Don't blindly believe everything you hear.
<Kolev>nckx, so there's a way to get swaylock to work without modifying config.scm now?
<nckx>Hell if I know, I don't even have a working Sway, but making the Guix swaylock package setuid is a dead-end.
<nckx>I do have swaylock configured but it's been months since I've locked my screen, it's something I never found a use for. It worked then.
<nckx>If the snippet in the manual is really no longer working, that deserves a buggy report.
<nckx>ACTION -> zzz
<nckx>To be clear, I'm referring to 'For example, swaylock fails to execute when compiled with PAM support and setuid enabled.' in case your manual's older.
<nckx>ACTION -> zzz!
<PotentialUser-32>Hey all, I've not been able to remove caps for ctrl despite having (operating-system (keyboard-layout (keyboard-layout "us" #:options '("ctrl:nocaps")))) set. I suspect xorg is overriding this setting somewhere. Anyone else having this issue?
<PotentialUser-32>guix system reconfigure was able to run correctly, I did reset my computer, working on getting a pastebin up of my config.scm
<PotentialUser-32>full config at pastebin dot com slash J4i7pNNn (not sure what the rules are about posting full links here)
<theesm>Kolev: you probably want to use screen-locker-service-type for swaylock =>
<podiki>PotentialUser-32: have to run, but you can post paste links. i also have swap ctrl/caps in my config but also in my xmodmap, and might only be the latter that works? will have to investigate
<podiki>ACTION off for now
<Kolev>theesm, thanks!
<PotentialUser-32>thanks, I have no xmodmap file, but do have setxkbmap installed globally. I'll look into putting it there
<Kolev>theesm, (name "swaylock-effects")?
<apteryx>nckx: you can also use 'mumi send-email *.patch' after outputing them with 'git format-patch'
<apteryx>mumi will poll Debbugs for the new issue and send the rest there
<apteryx>I've had mishaps using it though. not sure who's fault it was (mine, or a bug in mumi)
<mucho_maas>Hello. I'm new to guix. I'm getting a build error on `guix pull`. I've used `--keep-failed` but the directory in `/tmp` is empty.
<mucho_maas>What are the steps to further debug it? It's an error building `guix-package-cache`
<mucho_maas>In unknown file:
<mucho_maas>           0 (backtrace #<undefined>)
<mucho_maas>(exception unbound-variable (value #f) (value "Unbound variable: ~S") (value (python-brotli)) (value #f))
<nathan-web>I haven't been able to login to gnome since an upgrade yesterday. I get a lot of errors like: `gnome-session-binary[2774]: WARNING: Unable to find required component 'org.gnome.SettingsDaemon.XSettings'
<nathan-web>gnome-session-binary[2774]: DEBUG(+): fill: *** Done checking required components
<nathan-web>gnome-session-binary[2774]: DEBUG(+): fill: *** Session is not runnable`
<nathan-web>Any ideas on why these can't be found?
<nathan-web>When I log in via gdm, or via startx, I get an error like: "Oh no! Something has gone wrong. A problem has occurred and the system can't recover. Please logout and try again."
<adanska>Hi! Just asking here again: I'm having some trouble with reconfiguring my system - I'm trying to add an exfat file system called 'my-shared' that i intend to use to store files shared between my windows and guix installs. Heres what my file-systems field looks like, and the error, and the output of `ls /dev/disk/by-label`. I have exfatprogs installed on my profile, but not on my system profile yet. Would this affect anything?
<adanska>linux still has the capability to read and write to exfat file systems without exfatprogs, so i dont know why it wont work.
<adanska>nathan-web: are you able to roll back your system config? you could log into a tty using something like C-M-F2
<nathan-web>adanska: Maybe, but it would be a lot of changes to rollback. I was hoping to avoid it. I've been able to use a tty fine, so for now I'm going to just get another wm configured.
<adanska>ok. a bit worrying though. I wonder what commit caused it to break
<adanska>what commit was your previous generation?
<nathan-web>adanska: #5a6b1a
<adanska>nathan-web: oh wow thats a fair while back
<nathan-web>Yeah, Oct 15 or 16 I think
<efraim>Kolev: I have swaylock working on my machine, here's the relevant config bits:
<nathan-web>Is there any service like (set-xorg-configuration) for wayland? Specifically, I want an equivalent to `(xorg-configuration (keyboard-layout keyboard-layout))`
<adanska>i've found that gnome at least honours that.
<adanska>i've got that set and also the keyboard-layout field set and it works on my machine
<theesm>xkb keymaps should work with wayland as well => even though how a keymap is used/decided on is up to the compositor you use. Most should honor keyboard-layout settings afaik.
<nathan-web>Hmm, sddm doesn't seem to like the same configuration. If I use `(sddm-configuration (display-server "wayland") (xorg-configuration (keyboard-layout keyboard-layout)))` reconfigure gives me an error: `Wrong type to apply: #ubuntu<<keyboard-layout> name: "us" variant: #f options: ("ctrl:swapcaps")>`
<nathan-web>Err, fixing autocomplete: `Wrong type to apply: #<<keyboard-layout> name: "us" variant: #f options: ("ctrl:swapcaps")>`
<charlotte1123581>hi, i would like to make 'mk' from plan9ports available globally on my guix system. how can i get just the mk command from the package? the issue is plan9ports also overwrites 'ls' with the weird plan9 version for example, i dont want that. is there a reasonable way other than manually adjusting a bashrc alias to the /gnu/store every update?
<adanska>this is so odd... is exfat not supported on guix system or something???
<adanska>i mean, it is. I can mount an exfat file system
<adanska>it just doesnt make any sense why i can define it as a file system in my config.
<gabber`>i want to run a bash script with end-to-end tests after i install a package - but i fail. what am i doing wrong?
<gabber`>nvm i think i got it
<nckx>Guix System doesn't support exfat yet.
<nckx>Mounting it by /dev name should work, I think, but I haven't tried.
<nckx>Here's the list of file system labels & UUIDs that Guix can read:
<nckx>From experience, adding exfat support would be quite easy.
<nckx>sneek: later tell adanska
<sneek>Will do.
<nckx>sneek: botsnack
<nckx>And indeed, you don't need exfatprogs to mount it. That's all done by the kernel.
<gabber`>this: throws an ERROR: In procedure %resolve-variable: (expecting module): #f. any ideas what i am doing wrong?
<nckx>Missing #:modules (append %python-build-system-modules ...) or equivalent?
<gabber`>lemme check
<gabber`>#:modules as an list item in the (arguments ..) package field, right?
<gabber`>now i get "no code for module (guix build python-build-system)
<nckx>I'm not at a computer, but I'd also/first use the prefix:-style as used by gnome.scm gtg package, e.g. (assoc-ref python:%standard-phases 'foo) rather than the unstylish (@ ...).
<nckx>nextcloud-client is a better example, including #:importing the modules.
<gabber`>thanks! though i have some questions WRT you being able to post here without being at a puter ;)
<nckx>Whoever called it a 'supercomputer in your pocket' was tripping on crack.
<nckx>It's just extremely inefficient for tech support.
<nckx>And the support isn't as good.
<gabber`>i will!
<nckx>(What I meant is that I'd usually write up a complete example snippet, but perhaps this DIY approach is more pedagogical. Yes. That's the excuse I'll use.)
<gabber`>nckx: yeahyeah, i know you (at least that much) - and i learn OH SO MUCH rn. but do i really have to import all kinds of modules (e.g. (guix gexp) (guix store) (guix utils) and others) in #:imported-modules to make that work? it feels wrong
<gabber`>Lumine: o/
<Lumine>I eliminated all of my problems save for one in Guix over the past few days \o/
<Lumine>My home profile is much more sane now
<gabber`>ACTION wishes to be able to say something like that one day
<Lumine>The only problem that crops up is malformed URL in `guix pull' after using `guix gc' and a borked icecat because of it. Both problems are solved by clearing caches. jpoiret seems to think the culprit is libgit2.
<nckx>gabber`: No, that doesn't sound right.
<nckx>Especially that (guix gexp) & store is a red flag.
<nckx>Actually, they all are, importing (guix utils) on the build side is no-no time.
<nckx>ACTION away though.
<efraim>Apparently we still have files in the codebase which are iso-8859-1 encoded
<apteryx>janneke: hi! a recent commit of yours broke mes on i686-linux:
<apteryx>jlicht: the node-lts update broke ublock-origin-chromium ( could you look into it?
<elevenkb>Hello there, how might one see the standard output of a process that is running as a shepherd service?
<apteryx>it should be logged to /var/log/messages, I think, or to its own log file if the service defines one
<elevenkb>ok thanks, also is it possible to see the generated shepherd configuration file?
<apteryx>I usually do: guix gc -R /run/current-system | grep shepherd
<apteryx>then open in emacs with emacs-guix installed and M-x guix-scheme-mode
<apteryx>to have a readable version
<apteryx>what's the nix features that checks if a store output is unchanged between derivation called?
<apteryx>to avoid changing the graph is the output is identitical?
<jdek>does guix allow some form of 'system' dev libraries? I'm using nixos at the moment and it's a pain to do dev on random projects because I have to first setup a shell.nix etc
<janneke>apteryx: thanks for the heads-up, i'll have a loo
<apteryx>fun, teams (that video calling thingy) crashes in ungoogled-chromium
<gabber`>jdek: not sure what you mean *exactly*, but you can spin up and enter a shell with any package present like: $(guix shell glibc)
<jdek>If I have glibc globally installed i.e. in the default profile do I get the dev libraries too?
<apteryx>better yet, it crashes only when the conference starts, not letting you know earlier it`s about to crap out
<gabber`>apteryx: with f-fox conference calls work but p2p calls won't (unless you change your UserAgent to some Chrome version)
<gabber`>jdek: help me out here, the difference is the presence of the headers files in a /include directory?
<nckx>I too cannot parse 'dev libraries'.
<nckx>So: how do they differ from regular libraries?
<gabber`>now i'm running into "invalid syntax" in the wrap-program generated file... hmmm
<jdek>I think the main difference is that 'dev' libraries include headers somewhere accessible to the compiler by default, where by default can include pkg-config setting it up. Lets say I need sqlite3, I want to install sqlite3 to my system, and then not have to enter a guix shell in order to run a configure script and have that correctly find the lib and the headers
<jdek>nixos hides headers from the environment by default and you can only access them inside a shell which is what I'm trying to avoid
<gabber`>in guix terms you'd usually not want to have sqlite3 installed on your "system" aka *globally* (but maybe in your development profile)
<gabber`>what i usually do when i try to hack on and compile some package i enter a development shell specifically for that package with $(guix shell -D that_package) -- which ensures the package inputs to be readily available
<nckx>Oh. We don't hide headers or anything silly like that, although setting search paths is predicated on having a consumer like GCC present.
<janneke>ACTION tries to add m2-planet-1.10.0, but gets
<janneke>/gnu/store/y3vdq2pdkljrw63xxnc2vb6lz07ycar6-git-minimal-2.41.0/libexec/git-core/git-submodule: line 7: basename: command not found
<janneke>also, sed: command not found
<nckx>If Nix allows things in a shell that it doesn't allow in any other profile, including your user profile, then that's magic that Guix thankfully lacks.
<janneke>while building the git checkout; ideas?
<nckx>No, but it happened to me last week or so.
<jdek>it doesn't 'allow' per se, imo it's a bug that headers arent available by default but apparently its a design choice
<nckx>I didn't fix or investigate, my SSD gracefully destroyed itself to save me.
<jdek>but thanks, I'll definitely try out guix in a vm then
<nckx>jdek: Well, the Guix response would be 'even if you have GCC installed!?' but Nix might work differently.
<nckx>(Do they have search-paths? I don't even remember -_-)
<jdek>they wrap all their compilers, it's honestly a bit weird
<jdek>but yes even if you have a compiler installed in your default profile the headers dont show up outside of a nix-shell
<nckx>OK, I think we can safely say that Guix offers that killer feature, unless you're about to hit some unanticipated edge case later... :-)
<nckx>gabber`: Something is trying to [do the equivalent of] 'python'.
<nckx>Sorry if you already knew that, but there is no general answer beyond 'find out who and why and stop it'.
<nckx>It happens when something you thought was (only) a script is loaded as a module.
<jdek>can I patch packages without maintaining a branch of the entire guix repo?
<gabber`>jdek: you can have your own definitions locally (and even inherit the fields you want from available definitions)
<gabber`>nckx: nice, thanks! yes, i was doing that (by mistake). now i'm half a step further and get "/gnu/store/<hash>-foo/bin/.foo-real: cannot execute: Permisssion denied" and return code 126
<gabber`>it is defined in exec! aha!
<jdek>perfect, sounds like guix might fix my problems with nixos. (nix language, easy dev mostly)
<mirai>janneke: I think apteryx did some changes to it in core-updates
<podiki>howdy guixers
<podiki>churning away building mesa-updates; still a lot to go but if anyone has any favorite packages to check, take a look on CI
<RavenJoad>Just a quick check, is /tmp present in a build environment?
<podiki>caught one unexpected failure of gnome/gtk4 stuff due to a typo in upstream libdrm (which was updated on mesa-updates)
<janneke>apteryx: i've re-added m2-planet-1.9.0 and have mes use it; should be fixed thanks, but also...
<podiki>anyone know why a build will show as failed even though last line in log is that it succeeded??
<podiki>btw, pretty logs on cuirass are nice!
<podiki>and mbedtls-apache is failing some tests though nothing changed there but ungraft (of libx11) i think...
<old>where are the log of sshd?
<old>I can not find them under /var/log
<janneke>mirai: that's great...but it's broken on master right now
<podiki>janneke: regarding curl tests skip on Hurd, here's where I ungrafted and then added parallel test support, hopefully I didn't make a mistake: and
<gabber`>old: aren't they included in /var/log/messages
<podiki>janneke: I meant that the TFLAGS where a test has a ~ means something specific (i forget) and wasn't sure if that was important so i kept it (as opposed to straight test skip)
<old>gabber: right I just saw this
<old>anyway to see /var/log/messages other than sudo cat /var/log/messages ?
<janneke>podiki: OK
<janneke>i believe adding to disabled could work, but someone would need to check that
<gabber`>old: what ways are you thinking? there's `more' and `less' and all other kinds of inspecting text files
<old>gabber`: idk. Maybe there is something like systemctl journalctl?
<old>I don't mind the cat tho
<podiki>janneke: right, i left it as is to try not to introduce new changes (but parallel tests was such a huge speed up I added that)
<RavenJoad>I guess I answered my own question. /tmp IS available (as a name at least) to use in a build environment.
<nckx>I think the build is guaranteed to happen in the container's /tmp, even, regardless of the real TMPDIR.
<gabber`>old: not on Guix System (and that's a feature, not a bug ;)
<gabber`>do wrapped scripts (i.e. the .foo-real files) have to be set executable manually?
<gabber`>yes, indeed that seems to be the case...
<gabber`>how odd
<gabber`>(i guess i'm still doing it wrong though)
<nckx>I think they inherit/maintain their original mode.
<gabber`>i'm about 35% sure they don't (since otherwise the x bit would be set in my case)
<nckx>How sure are you that the original really has that mode, and that it's not being stripped elsewhere? Let's keep adding variables until the odds round down to 0.
<nckx>Alternatively, if you can 'prove' with 2 or 3 strategically placed (invoke "ls" "-l")s that this really is happening, that would be cool too.
<nckx>ACTION surprised.
<gabber`>nckx: i do the $(ls -l) and WHOA: there it is
<gabber`>my package suddenly started to work after i chmod +x'd the .foo-real file
<nckx>Oh I believe you, I'm just not convinced that wrap-program's to blame for its absence.
<gabber`>i should add that - as a person who probably will never play russian roulette - 35% certainty with me is quite a lot
<nckx>Let us never go drinking.
<nckx>Let us for that matter never go to Russia.
<gabber`>it probably isn't (i have my copy-build-system borrowing the python 'wrap phase) -- but i couldn't find (obvious) clues on where else that should happen
<gabber`>nckx: i also rarely say never (except for, maybe, just now)
<nckx>I'd normally ask for the package (here or in PM) but I'm not in a position to interrogate it myself.
<gabber`>i'd rather not share (it's on the ugly side leaning on forbidden with a note of cursed)
<gabber`>i might provide a dummy-test example if i'm in the mood tomorrow
<nckx>I can respect that.
<nckx>Speaking of cursed, I'm starting to suspect/realise (not sure yet) that exFAT doesn't have fixed locations for volume labels & GUIDs, or even trivially indirected ones.
<nckx>Long shot but if anyone's familiar with the format, er, hit me up.
<mirai>gabber: I think the _original_ file has to be executable in the first place
<mirai>after all, wrap-program is a script that executes .foo-real
<cartographic>Oh man guix is throwing errors on the command line (Debian). Any ideas?
<mirai>and wrap-program renames the original file to .foo-real
<apteryx>re nix features that would avoid full rebuilds on same-output/different-derivation changes: the "intensional store" RFC:
<PotentialUser-83>Hello, #guix, I was just trying to run the chickadee example, but unfortunately I get the following error:
<PotentialUser-83>ice-9/boot-9.scm:3330:6: In procedure resolve-interface:
<PotentialUser-83>no code for module (chickadee cli)
<jpoiret>cartographic: do you get errors with `LC_ALL="en_US.UTF-8" guix`?
<jpoiret>PotentialUser-83: how are you running this? are you using `guix shell` or similar?
<PotentialUser-83>No I'm using the GuixSD and I'm just running `chickadee play example.scm`
<cartographic>jpoiret: thank you. Do you mean to quite litteraly run $ `LC_ALL="en_US.UTF-8" guix` ?
<cartographic>If so, i get guix: missing command name
<jpoiret>PotentialUser-83: did you install `guile` and `chickadee`?
<jpoiret>cartographic: at least it's not crashing then :)
<PotentialUser-83>jpoiret, yes I did. I'm using Guix System Distribution.
<jpoiret>seems like there's a problem with your locales 🤔 have you `guix pull`d yet?
<jpoiret>guix should use its own locales, and so Debian's shouldn't interfere
<cartographic>If only - running any guix command returns the same error :/
<cartographic>`guix pull`, `guix --help`, `guix package ...`
<jpoiret>right. Can you try `LC_ALL="en_US.UTF-8" guix pull`? (without the backquotes)
<jpoiret>the Debian-provided guix not working is often a Debian problem
<cartographic>That works :)
<jpoiret>once you've pulled, the new Guix should not exhibit the same problem
<cartographic>Ace. thank you!
<cartographic>jpoiret: Thanks - after the pull it still errored but setting LC_ALL in my .bash_profile fixed this.
<jpoiret>cartographic: that's problematic though
<jpoiret>do you have a specific locale usually?
<jpoiret>the LC_ALL environment variable will apply to literally *all* of your programs, Debian's included
<cartographic>Hmm. en_GB.UTF-8?
<cartographic>jpoiret: to be honest, it's very much past the time I should have left the office, so I think I'll pick this up tomorrow
<cartographic>\Looks like I should do something with locale instead
<cartographic>Thanks for your help - night all
<gx1>having some trouble getting docker to work, just trying to get 'sudo dockerd' to run (not messing with getting shepherd to start it automatically yet), but I'm getting 'failed to start daemon, Devices cgroup isn't mounted'. I haven't messed with cgroups, but haven't had this issue with dockerd on other distros. Anyone run into this?
<gx1>this is docker on guix SD btw, trying to run a docker-compose stack
<zamfofex>gx1: I think it will be easier to set up the daemon with the Shepherd service.
<zamfofex>It should get everything right for you automatically.
<gx1>is there a snippet somewhere which could help me do this without too much knowledge? I'm still new to shepherd and guix in general
<gx1>or maybe a service type I could add to my config.scm
<zamfofex>You should be able to just add ‘(service (docker-service-type))’ to your services, reconfigure the system and reboot.
<zamfofex>You can add configuration if you’d like, but it’s optional (it has sensible defaults).
<gx1>I see, working on it now. Seems like I need elogind, and it's not enough to just add that to the packages list in config.scm
<gx1>is that another service type? I did a quick ctrl-f on the services in the manual there
<zamfofex>Ah, yes. ‘(service elogind-service-type)’
<zamfofex>I feel like it should do that automatically for you. (Most people have an elogind already, because it is used by e.g. GNOME or so, I think, so I imagine it might be easy to miss the fact Docker depends on it.) If you’d like, you could file an issue about it.
<zamfofex>Also, not many people use Docker here, I think.
<gx1>I'll consider it, although I'm not sure I know enough about what I'm doing to really even articulate the ticket beyond 'this is a dependency of docker and should probably be pulled in when the docker service is enabled'. I've kind of been flying by the seat of my pants
<gx1>What do people here use instead of docker? Just guix containers? I find it useful for selfhosting / making my own images and sometimes running something without using npm
<zamfofex>Yeah, Guix shell containers when necessary.