IRC channel logs


back to list of logs

<graywolf>Since it is quiet here, totally off topic question: I would like to buy some local products in french eshop, but they ship (surprise) only to France. Would there be someone living in France willing to receive the package and sent it over to Czech republic (so within the EU). Not for free of course. If so, please drop me a private message.
<graywolf>Again sorry for the offtopic, but I learned that Guix is French Nix and I failed to find French shippito :/
<lechner>graywolf / i would help you but do not live in France
<Kolev>How to get I2P on Guix System?
<morte>Kolev: I use i2pd, I haven't tried to setup a custom tunnel
<morte>but browsing works just fine
<Kolev>morte, just keep it running in a shell somewhere?
<morte>yup, I use tmux, so I just leave it there when I need to
<Kolev>morte, what do I run on the command line? Do I need to configure my system in any way?
<morte>Kolev: ok, so you install i2pd via the guix command, then run i2pd in a terminal or tty, once it's running it opens a http proxy on so you just point your browser's proxy to that address and you should be able to browse any i2p sites
<morte>I just did it with lua kit and it's working fine
<graywolf>Can someone try to build libtorrent-rasterbar (I want to verify if the failure is only on my end). guix build --max-silent-time=600 --check libtorrent-rasterbar
<graywolf>(For me the build just hangs, therefore the --max-silent-time)
<Kolev>morte, 4444. Got it.
<podiki>graywolf: times out on ci as well: probably needs a test disabled (max silent time there is 3600s I think, based on log)
<peanuts>"Build 2525764"
<podiki>it is rather out of date according to guix refresh as well
<Gooberpatrol66>coreutils is located in gnu/packages/base, i added the gnu/packages/base module, and added coreutils as an input, but it still tells me "exception unbound-variable (value (coreutils))"
<Gooberpatrol66>is there a way to rename a file in a build without "invoke mv"?
<dcunit3d>does anyone know any emacs-xyz packages built with eldev?
<dcunit3d>i'm trying to fix this issue with emacs-citar. the autoload validations should fail for this
<peanuts>"branch master updated: gnu: emacs-citar: Fix build."
<dcunit3d>it's basically this issue, but it's pretty esoteric for me
<peanuts>"Re: autoload cl-defstruct constructor?"
<lechner>packages.ggo is not responding to HTTPS requests
<the_tubular>What's that lechner ?
<the_tubular>Ohh, my bad lol
<the_tubular>Though ggo was a tld I've never heard of
<lechner>i should have written packages.g.g.o
<lechner>it is customary at debian to write things like tracker.d.o but i'm to lazy for the dots, too!
<lechner>two times too
<the_tubular>4 ?
<the_tubular>Sorry :P
<lechner>that's funny
<bumble>hey everyone I wish someone will take a look at these ones
<peanuts>"[PATCH]: Add PyQt 6."
<peanuts>"[PATCH] gnu: qutebrowser: Update to 3.0.2."
<isaneran>when running a really big guix gc
<isaneran>it prints that it deletes maybe 20-30 items
<isaneran>then it stops for a bit
<isaneran>and repeats
<isaneran>is this Mr Boehm GC interrupting the flow or why does this happen?
<lechner>how big a gc?
<isaneran>up in 40 GB so far
<isaneran>ok maybe 20-30 is wrong, more like 100
<isaneran>but it does have a very distinct start stop behaviour, instead of like a steady stream of oh deleted this and this and this
<lechner>i think that's kind of common
<isaneran>it's more like deleting THISTHISTHISTHSI... wait.... THISTHISTHIS
<isaneran>yeah I don't think it is an issue
<isaneran>I was just wondering why it happens
<lechner>i just ran a gc with 350 GB on monday. it takes time
<isaneran>and suspected it's maybe guile stopping to gc itself
<lechner>probably some symlink traversal
<isaneran>I guess maybe, but I feel like the start stop pattern would be less regular if it was file system related
<lechner>profiles tend to be similar
<isaneran>can imagine 350 taking time yeah, mine was just 54 in the end
<lechner>especially system profiles
<isaneran>I think it stopped many more times than I deleted profiles though hmm
<vivien>lilyp, from what I understand, gnome-bluetooth is indeed responsible for libadwaita in the gnome-shell inputs, but kicking libadwaita out does not solve the build failure. Gnome-shell still has gtk in its inputs instead of gtk+ but I don’t get why it is a problem *now* and not a few weeks ago
<lilyp>1c41971e721dde203580ec17899beae546f1133a already appears bad imho; prior to the merge
<peanuts>"guix.git - GNU Guix and GNU Guix System"
<lilyp>We sadly don't have guix graph for propagations
<lechner>Hi, is gnunet building for anyone? Locally, it seems to be stuck in the 'check' phase
<lechner>actually, maybe not. it's just exceptionally slow
<lilyp>vivien: docbook-xsl
<isaneran>hum_equiv: puff
<adanska>Hi Guix!
<adanska>Hi isaneran! How are you?
<isaneran>pretty good!
<adanska>pretty good too. just doing some work at the moment, but i've also been procrastinating by having a look in on how the upgrade to GNOME 44 has been going
<adanska>vivien has been absolutely leading the charge! theyre doing so much work, its very impressive! i just wish there were more people that could review their work, i'm not experienced enough to contribute much to the conversation at this point ;(
<lilyp>At least it's a two girl show rn; it used to be just me doing the work and reviewing it :)
<isaneran>unfortunately can't help out, not good enough at guix and also not using gnome so there is less of a dogfood factor
<adanska>very fair
<isaneran>but it's important work so I applaud the effort
<adanska>i'm using gnome at the moment, but i'm still not quite experienced with guix and GNOME apps to make any sort of meaningful comments/contributions. but! its the uni holidays, so maybe that could change with a bit of perseverance haha
<isaneran>take it little by little
<isaneran>see what parts you understand, which ones you don't, check manual, ask etc
<adanska>absolutely. i dont want to overwhelm myself.
<isaneran>and don't put too short of a time horizon on getting fluent
<adanska>yeah, thats good advice. i think i'm going to try something completely different by implementing a home-onedrive-service of some kind. we already have a onedrive client packaged, so i just need to figure out a way to design the config records, dealing with multiple accounts etc
<adanska>it has that dogfood factor too since i use it for university work
<adanska>the worry i have is storing the authentication token securely using something like gpg. ive heard that storing sensitive data for use with guix services can be pretty tricky since the store stores it in plaintext
<lilyp>yeah, ya can't store secrets
<lilyp>if you'd need to do so, hold off on writing your service
<adanska>really? that much of a showstopper?
<lilyp>if you want to contribute to gnome-team, there's probably some gnome-adjacent apps that'd like to be upgraded
<adanska>nice! ill have a look into it :)
<lilyp>adanska: depends; if you have to specify a password in the config file itself, that's a big nope
<isaneran>authenticator doesn't start on my machine so maybe that's one :P
<isaneran>but i ended up using some emacs totp thing
<lilyp>if it uses libsecret or anything else and just does a key lookup, then that's not sensitive data imho
<isaneran>lilyp: you could specify a function for retrieving the password though, and that function could be doing all kinds of things like just returning it, calling a command to retrieve it, or reading a file to retrieve it etc
<adanska>lilyp: yeah, the onedrive client currently just stores the access token in a plaintext file. pretty insecure, but i think there may have been a way to use something like libsecret to make it better in the docs
<civodul>Hello Guix!
<lilyp>and then there's mega projects like porting guix to webassembly to have it run in your browser :)
<isaneran>Hey Guik!
<adanska>lilyp: damn, i've never heard of that! that does sound mega indeed
<adanska>is it still being actively worked on?
<adanska>hi ludo', efraim!
<efraim>adanska: that'd be amazing if you got a home onedrive service going. I forget why I packaged it.
<efraim>you could probably even start with just a start/stop service and have people enter their credentials in ~/.config/onedrive/config or wherever it normally is stored
<adanska>yeah, that would be a good start! then work on migrating all the config fields to a config record.
<lilyp>vivien: running gnome-shell in a post-fail checkout, the perf tests all seem to time out
<lilyp>and it's due to dbus ._.
<adanska>._. dbus ._.
<isaneran>.-. subd .-.
<efraim> here's what I currently have locally for a onedrive service. Never tested it, but it works for my dbxfs service, which is just above it
<peanuts>"~efraim/guix-config: efraim-home.scm - sourcehut git"
<adanska>when a test fails in guix, 9/10 its dbus, the other times its something else that just ends up being dbus :p
<adanska>efraim: thanks! ill have a look :)
<adanska>a priori always helps
<isaneran>eye health PSA: remember to take frequent breaks to blink and look far away. When you use the computer it's easy to forget to blink, and your eye gets dry and it becomes easier to get an infection
<isaneran>and probably other issues
<lilyp>well, sure, but that's because dbus is pretty much everywhere
<efraim>did at-spi2-core get its 'check phase to honor #:tests in the gnome-team branch?
<lilyp>not yet, will have to merge master
<adanska>isaneran: 👁️ you've just triggered manual blinking 👁️
<isaneran>'pparently you wanna blink pretty hard
<isaneran>not always, but in the break if you're only able to take like a 10-20 second break
<adanska>really? well, im going to go take a break now and i WILL be blinking thoroughly and with ample force
<lilyp>I always blink subconsciously, I don't have time for your mind games.
<adanska>might do a lunge or two, if im feeling limber
<isaneran>yeah might as well combine it with other things you should be doing in your break anyway haha
<isaneran>but try to look 6m away at least
<isaneran>lilyp: I was relying on my subconscious blinking too but then my eyelid got swole and I got lectured by mr doctor
<adanska>ACTION shall be away for 10 mins whilst they blink and lunge 
<isaneran>ah if you're away a full 10 minutes you probably don't need to blink that much (assuming you're not looking at the phone etc). Though it's not a bad ideato do it
<adanska>haha no it was not jam packed full of blinking i got some snacks n water
<adanska>i've sorta inferred what it means, but what does the % prefix naming convention mean in scheme?
<civodul>adanska: it’s generally used for global/immutable/“important” variables in Guix and Guile
<peanuts>"A Scheme Crash Course (GNU Guix Cookbook)"
<adanska>ah, cool. thanks!
<adanska>i really like how you've laid out your guix config, efraim. very inspiring! i might try to adopt a similar organisational structure...
<efraim>thanks. I'm think I should've alphabetized the config files to make them easier to find but otherwise I more or less like what I've done with it
<adanska>i really like how it leverages guile to dynamically compose the config. its very nice. i dont have much of a usecase at the moment for something like that, since i just have my laptop. however, if i ever need to manage more machines with guix-system/guix-foriegn, now i know how!
<civodul>oh mesa-updates merged, congrats podiki, efraim, and all!
<civodul>looks like mesa-opencl fails to build:
<peanuts>"Build log of build #328518"
<civodul>does that ring a bell?
<efraim>I thought it got confused between llvm-for-mesa and llvm but I'm not so sure
<efraim>and the source makes it look like llvm-15 is plenty
<efraim>i'll take a look at it
<zilti>efraim: How did you get file-system-label to work? For me, guix always complains that there is no device with that label
<efraim>for the rust-team branch I'm now really just waiting for the build machines to catch up
<efraim>I prefer UUID, but the SBC guix image builder templates set the file system label so that normally works for me. or more likely u-boot makes it work
<civodul>rust-team is at 70% at
<civodul>did you rebase it on current master?
<adanska>zilti: what filesystem is your parition using? some file systems such as exfat simply dont work in guix yet, although someone sent me a patch that implements it, i havent gotten around to testing it yet
<efraim>civodul: yeah. but wait for to come back up. substitutes for aarch64 is way too far behind
<zilti>adanska: the installer default for the EFI partition, and xfs for the root
<zilti>I did set the labels in the TUI installer
<adanska>odd. doest /dev/disk/by-label show your disk?
<zilti>I only have /dev/disk/by-partlabel
<adanska>hmm.. that is very odd.
<adanska>im not very familiar with xfs so im not sure where to go from here unfortunately. maybe try and relabel the parition using something like gparted?
<zilti>So I suppose the installer only sets partlabels
<zilti>I'll try that, yes
<civodul>ACTION considers submitting a talk about Shepherd for
<efraim>I was thinking my WIP blog post would be a good one for the distro room
<civodul>oh yes!
<civodul>i think we never had a talk in the “container” devroom, but it’d be nice too
<civodul>just presenting what we already have
<efraim>also I ran into some oddities with the firmware for qemu, thought that might be good for the openfirmware/bmc/bootloader room
<civodul>sounds nice too
<civodul>reaching out to other communities is always pleasant IME
<civodul>another one on my list is the SBOM devroom
<efraim>I have to reach out to the tow-boot people, I think they'd be great for the declarative & minimalistic devroom
<civodul>ACTION doesn’t know tow-boot (yet)
<efraim>it takes u-boot, adds some nicities and packages it up for SBCs like the pinebook pro. whole thing is managed using nix
<efraim>so it gets declarative AND minimalistic
<zilti>Also, I wonder why there is no partition-label function
<gabber`>zilti: there is, in (gnu image)
<gabber`>as a part of the record-type <partition>
<zilti>gabber`: but can I use it to declare (file-system (declare (partition-label "guix-root")))?
<gabber`>zilti: inside an operating-system definition?
<gabber`>i don't think the operating-system definition sets up formatting, partitioning and labelling of disks
<zilti>It does not, but it'd be nice to use partition-label inside the `file-systems` definition for the mounts
<gabber`>there is this find-partition-by-label in (gnu build file-systems)
<zilti>I tried that one, too, but that only returned #f for my partitions :/
<gabber`>have you had a look at the reference manual section for "File Systems" (in "System Configuration")?
<zilti>I have, yes, and it only mentions `uuid` and `file-system-label` - or using a /dev/... string. So I of course also tried `(device "/dev/disk/by-partlabel/guix-root")`, which also did not work
<efraim>civodul: and are back up
<peanuts>"Branch master Guix Quality Assurance"
<peanuts>"Branch rust-team Guix Quality Assurance"
<civodul>efraim: the differences between the two build farms are impressive
<efraim>someone tell nckx to pick up their game, they're way behind on substitutes and architectures
<gabber`>can i start a db (postgres) in the build phase of a package and connect through it? is the /tmp from inside the build process the same as on my host system?
<civodul>gabber`: i would add a pre-check build phase and start pg_tmp
<civodul>(from the ‘ephemeralpg’ package)
<graywolf>Is anywhere documented how does the guix build environmet look like? Or even better, is it possible to get a shell inside it? I have a test that fails during guix build, but passes in guix shell -C, so I am trying to determine the cause...
<singpolyma>You can use -K to keep the temp dir
<singpolyma>Then source the environment variable file in there and run commands mostly rs if in build env
<graywolf>Will that also take care of isolating the network and such?
<singpolyma>But guix shell --pure -C should? Then do it inside there?
<graywolf>Hm, as a starting point I can try to run the tests in tem dir by guix shell -C,
<graywolf>Yeah I had the same idea
<graywolf>Will try
<graywolf>btw do I need to use --pure with -C?
<graywolf>I always to just -C, is that a mistake?
<singpolyma>I'm not sure, with -C maybe they're the same
<civodul>graywolf: ‘--pure’ has no effect when using ‘-C’
<civodul>perhaps it should print a warning
<civodul>‘-C’ is like an enhanced ‘--pure’
<graywolf>Oh, -K prevents offloading?
<graywolf>In that case I guess I have time for lunch :D
<civodul>building webkitgtk on your laptop? :-)
<gabber`>civodul: thanks for the hint!
<sneek>yewscion: wb!!
<yewscion>sneek: botsnack
<zilti>How does one use greetd-wlgreet-session? Especially, how do I set the `command`? It accepts neither a string nor a `file-append`, in both cases it fails with "Invalid entry of a G-Expression"
<graywolf>lol.... ok, so the reason for libtorrent-rasterbar's test failures (well, timeouts) is `faketime'.
<graywolf>Well this was "fun".
<attila_lendvai>is there a strong emotion against using capital letters in format strings, i.e. ~A, ~S, etc? (i.e. will the patch be rejected?) i find them much more readable.
<zilti>I went into the source code, and even when I try to put the default there, so, `(command (file-append sway "/bin/sway"))`, it fails with the same error
<sneek>wb yewscion
<yewscion>sneek: botsnack
<apteryx>weird, the ethernet port on a X200 doesn't seem to work with Guix
<apteryx>is this known?
<mrvdb>hi guix, is anyone working on updating the telegram-desktop package?
<zilti>The reason why I am using it in the first place instead of `greetd-wlgreet-sway-session` is that the latter is broken in that it ignores if one sets a different package as the sway package
<graywolf>What is coding style for changelog when I want to do `* file (foo)[bar]' and `* file (foo)[baz]'? I know that I should not repeat the file, but I do not know if it should be `* file (foo)[bar]' and `[baz]' or `* file (foo)[bar]' and `(foo)[baz]'.
<graywolf>Could any commiter advice please?
<jpoiret>just put [baz]
<graywolf>jpoiret: thanks, one more: for arguments, the pattern is [arguments]{#:phases}<'what-phase> ?
<jpoiret>i've never used something this deep in changelog format unfortunately.
<jpoiret>i'd just look at what's been used in the log
<apteryx>graywolf: there's no strict format; we're already stretching the style according to 'info standards'
<lilyp>you typically want [arguments]<#:phases> and that's about it
<graywolf>I see. Well, based on git log I will go with [arguments]<#:phases>[what-phase]
<lilyp>if you don't have other phases to frob, use [#:phases] to save a level
<apteryx>but [phases] <the-phase> may be acceptable as a shortcut (I think [phases] {the-phase}: was used in the past)
<graywolf>Got it. I am touching only 'check, so will leave out the final level
<apteryx>more correct would be [arguments] <phases> {the-phase}, but this is getting deeper than documented in 'info standards', so the { } is custom
<graywolf>my 2¢: I get that "these's no strict format" but it would still be nice if there were
<graywolf>So I could just follow the info manual
<apteryx>and even then [ ] (brackets) are supposed to be for conditionals, not for records fields (so as I've said, we already have our own dialect)
<sneek>wb yewscion
<apteryx>does no-one here uses a x200 connected via wired ethernet?
<graywolf>Released in 2008? Nope, sorry...
<apteryx>it's GNU Boot compatible ;-)
<aldum>ethernet doesn't work?
<apteryx>not on mine
<apteryx>so I'b
<aldum>same issue here
<apteryx>oh! so it's confirmed
<apteryx>I'll research if it's a (non-free) firmware issue
<apteryx>That'd be the first Ethernet chipset I've encountered that doesn't work using free software
<apteryx>aldum: thanks for the info!
<aldum>I don't think so, it doesn't work with nonguix either
<apteryx>hm. I wonder what it might be then.
<aldum>or rather, (if I recall correctly) it did work with the guix installer, but doesn't on the installed system
<aldum>I haven't had time for this in a couple months
<apteryx>the TP150 USB adapter I had been using has recently become unreliable, probably due to 2.4 GHz congestion around the house more than because of some failure (it connects and disconnects erratically)
<apteryx>so I tried wired ethernet, and was suprised it couldn't connect
<ieure>Looks like it's a PCIe Intel Ethernet on that, uses the e1000e driver.
<ieure>X200 is a great laptop, wish they made them more like that today.
<apteryx>I don't see any issues reported regarding wired ethernet on its h-node page:
<peanuts>"ThinkPad X200 (7458V2M) -"
<aldum>ieure: tell me about it
<aldum>I just tried to look at new laptops, the keyboards are still (or even more) atrocious
<dthompson>I had to get a new laptop last year. after a lot of research I ended up right back where I started: a thinkpad. yeah, they don't make em like they used to, but I needed modern hardware. it's been just about a year and I don't regret it.
<graywolf>Is it allowed to add nss-certs and openssl into propagated-inputs? Or should the user be expected to install them if deemed necessary?
<katco>dthompson: working well with guix i assume? might be in the market for a new laptop
<dthompson>katco: yup!
<ieure>katco, Hi. :)
<dthompson>my up-to-date config is public:
<graywolf>Hm, I guess the wrapper programs it the corrent way now that I think about it?
<peanuts>"ikaruga-os.scm - guix-config - My Guix System configs"
<zilti>...guess manually starting swayfx will have to do for the time being then ¯\_(ツ)_/¯
<katco>ACTION wonders if anyone's using guix successfully with the framework laptop
<ieure>If they ever get around to making a Framework with TrackPoint, I'll switch.
<katco>can't you swap the keyboards on those pretty easily? i wonder if there's a mod :p
<peanuts>"Finally, a TrackPoint Framework - Mainboard - Framework Community"
<katco>XD there ya go
<dthompson>the frameworks look like they're getting pretty nice. I was recently at a conference and overhead some people talking about theirs and they said sleep mode doesn't work right on linux.
<katco>ACTION goes for a walk
<dthompson>that seemed to be their biggest complaint.
<aldum>ieure: exactly what I said
<aldum>there's hope, the new 16 inch one has input modules
<aldum>sleep mode never works right on linux :) but what was the complaint?
<dthompson>that's as much as I heard
<dthompson>my thinkpad has no issues with suspend
<dthompson>none that I can tell anyway. very little battery usage when the lid is closed.
<Rovanion>All my laptops have had 0 suspend issues the last 15 years, except this Thinkpad T14 with an AMD chip that only "Windows modern sleep".
<graywolf>Ok, one more question: Is it preferable to put openssl into propagated-inputs OR to set native-search-paths instead?
<lechner>sneek / later ask mirai / Hi, how should I deprecate krb5-configuration (which I renamed) please? Using define-deprecated gives an error. Thanks!
<sneek>Will do.
<apteryx>graywolf: native-search-paths
<graywolf>Thank you :)
<apteryx>aldum: re x200 problematic wired ethernet, interestingly, other users seem affected:
<peanuts>"Intel ethernet device on a Thinkpad X200 not recognized / Kernel & Hardware / Arch Linux Forums"
<lechner>Hi, why would 'git send-email' for an innocuous typo fix, add X-Debbugs-Cc: Christopher Baines <>, Josselin Poiret <>, Ludovic Courtès <>, Mathieu Othacehe <>, Ricardo Wurmus <>, Simon Tournier <>, Tobias Geerinckx-Rice <> please?
<attila_lendvai>lechner, see etc/teams.scm
<lechner>attila_lendvai / i did not copy any teams, though
<graywolf>lechner: the script does that by itself based on the content of the patches
<lechner>i fixed a single typo
<peanuts>"guix: deprecation.scm: Fix docstring typo. ? d6fec8eb13 - guix -"
<apteryx>lechner: the teams script works at the file level, nothing more elaborate
<apteryx>so if you touch a file under the supervision of the core team, it'll CC the core team
<lechner>okay, thanks!
<lechner>i sent it. just want to make sure those folks don't get into a tizzy
<lechner>Hi, is there any point to the --container option to guix shell when running Guix System? Thanks!
<graywolf>I use it to isolate rest of the system from the content of the container (shell) to make sure nothing is leaking in or out.
<lechner>graywolf / like process information?
<graywolf>guix shell -C does few (for me) interesting things: 1. Mount fake filesystem exposing only cwd and store (by default) 2. Disable network (unless -N) 3. Do not let in any environment variables.
<graywolf>1. is useful because when testing (for example) new version of a program, it does not see files at ~/.config
<Baptiste>Hello everyone, if there are any Emacs user here at the moment: has anyone been having trouble with Emacs and Tramp since its new update (29.1)? Namely with eshell. I cannot use `sudo` anymore nor can I use any command on a remote host. Does anyone relate to that?
<lechner>sneek / later tell mirai / nvm, i referred to the configuration in record syntax <> instead
<sneek>Will do.
<graywolf>Baptiste: I did have some issues that would fit, but assumed it was due to having a port in my remote directory path
<Baptiste>graywolf Huum, interesting. And I imagine your `sudo` doesn't work either? I either get a "must be owned by uid 0" or a "command not found"
<graywolf>Yeah, usually the command not found. It tried to run the remote commands locally I think.
<graywolf>I just gave up and used shell-mode
<graywolf>Very quick check was `id' giving my local user (I think, it is some time...)
<Baptiste>graywolf: Ok thank you. I'm going to file a bug report
<apteryx>aldum: this is my dmesg output, + ip link show enp0s25, which is my network device:
<peanuts>"debian Pastezone"
<ieure>Really wish peanuts didn't expand links for, they're not very useful.
<peanuts>ieure: Hi, for comments please contact my maintainers at
<Franciman>is there already someone working on packaging a newer pipewire version?
<aldum>if you have enp0s25, then it's there
<ieure>Yeah. Cable issue maybe? I see multiple link up/down messages in that dmesg output.
<aldum>so it's still not working?
<aldum>for me, it doesn't even show up
<apteryx>ieure: it had ping work for a brief moment
<apteryx>I've tried with the kernel argument pcie_aspm=off, and now it at least tries connecting repeatedly
<apteryx>maybe I could add debug to have more ideas about what is going on
<apteryx>I'll try going back to an older kernel, perhaps it's a regresion, this is happening with 6.3.13
<apteryx>from July 28 ^^'
<apteryx>may be a networkmanager thing
<apteryx>I should have checked /var/log/messages
<apteryx>same issue on 6.2.12 (April)
<apteryx>same all the way to the oldest 5.15.16 kernel I still have
<Franciman>how do i prevent pulseaudio from getting started?
<Franciman>i am not using any DE and i removed pulseaudio-service-type from %desktop-services
<Franciman>what do i have to remove?
<apteryx>you don't use applications that want it
<Franciman>oh that too, gorg. Thanks
<apteryx>there may be a config way to avoid them trying to use pulseaudio in the first place, i'm not sure
<apteryx>but otherwise since they were built with pulseaudio support they'll try to have it autospawned
<Franciman>ok it's unescapable
<Franciman>i use icedove which has pulseaudio as dependency
<Franciman>i wanted to use pipewire-pulse as a drop in replacement
<apteryx>you could use a custom icedove rewritten to have that dependency removed
<PotentialUser-70>how to install and use tkinter?
<apteryx>I think it's an output of the python package
<apteryx>guix install python:tk
<apteryx>or something
<Franciman>apteryx: lol yes, but it's a huge endeavour :P
<Franciman>i'll stay with pulseaudio haha
<ieure>Ugh, disappointing that Guix is stuck with Pulse.
<Franciman>well, actually you can do this trick
<jpoiret>ieure: it is not
<Franciman>you can modify pulseaudio's package to install and use pipewire :)
<jpoiret>as long as you started pipewire-pulse before pulseaudio you'll be fine
<jpoiret>i've been using pw on guix for >2 years now
<Franciman>jpoiret: do you have a config to share?
<apteryx>jpoiret: excellent bit of knowledge
<jpoiret>and yes you can disable PA's autospawn, with a system configuration file (you can also configure that through Guix :) )
<PotentialUser-70>apteryx its not working as an output of python package
<ieure>Okay: It's disappointing that Guix ships Pulse by default and you have to mess with stuff by hand to get rid of it.
<apteryx>PotentialUser-70: do you have python installed as well? 'guix shell python python:tk'
<jpoiret>Franciman: not really, i just have the following bit in my sway config
<PotentialUser-70>yes I have it
<peanuts>"debian Pastezone"
<apteryx>PotentialUser-70: what is the exact issue you are facing?
<jpoiret>ieure: i agree (esp. since pw is way better than pa), but widespread adoption by distros was quite recent
<jpoiret>and also it's only now that they've released 1.0!
<PotentialUser-70>I hav ean import of tkinter in a project, but I cant find how I can install it
<PotentialUser-70>Usualy this is a separate package, but there is no one in guix
<ieure>PotentialUser-70, It's a separate output of the python package. Here's a patch I submitted recently for a Python program that uses TK for its UI:
<peanuts>"[bug#67120] [PATCH v2 4/4] gnu: Add jellyfin-mpv-shim."
<Franciman>thanks jpoiret. So you start wireplumber before pipewire. Interesting
<ieure>PotentialUser-70, The relevant section there is `(,python . "tk") -- this is telling the system that the package recipe depends on the tk output of Python. There is no install process, you declare which things you need as inputs to your recipe, and the system makes them available.
<apteryx>seems like they're developping on a project, so I thought they'd prefer 'guix install' or 'guix shell'
<PotentialUser-70>I have installed python:tk but it still doesnt work, maybe because tk, as I found, is a tensorkit sometimes, not a tkinter
<jpoiret>Franciman: no, before pipewire-pulse, but pipewire itself is started before wireplumber
<apteryx>ok, yeah that's something entirely different
<jpoiret>you can see that I even wait for the pipe to be available
<lechner>sneek / later ask mirai / actually, deprecating records does not work gracefully. it errors out because the deprecated record does not exist. ("error: krb5-configuration: unbound variable") Is there a solution?
<Franciman>wonder whether i can make it work via a guix home service
<Franciman>would be grand
<sarg>Franciman: you need pipewire home service? It's in guix-patches, waiting to be merged:
<peanuts>"[PATCH] gnu: home: Add support for home-pipewire-service"
<ieure>Is it just me, or do patches seem to take a very long time to get looked at?
<sarg>pipewire 1.0.0 update is trivial, just add doxygen and python to native-inputs
<graywolf>ieure: It is not just you. Sometimes it is fast, but I am still waiting for review of patch from August :)
<ieure>Related, if someone wants to take a look at the LibreWolf browser stuff I contributed:
<peanuts>"[bug#67512] [PATCH 0/5] Add LibreWolf"
<PotentialUser-70>Does anybody have ideas about tkinter?
<ieure>PotentialUser-70, You need to provide more information to get any help. Such as a paste of the output showing what's not working.
<PotentialUser-70>well, it just says that no tkinter found     import _tkinter # If this fails your Python may not be configured for Tk
<PotentialUser-70>ModuleNotFoundError: No module named '_tkinter'
<ieure>PotentialUser-70, The module is 'tkinter', why's it trying to import '_tkinter'?
<peanuts>"tkinter ??? Python interface to Tcl/Tk Python 3.12.0 documentation"
<ieure>> [_tkinter] is automatically imported by the main tkinter module, and should never be used directly by application programmers.
<ieure>PotentialUser-70, In the absence of more detailed information, it looks to me like the thing you're trying to run is buggy.
<PotentialUser-70>No, its just a debugging message from the library, in the project it imports without _
<podiki>graywolf: what patch?
<ieure>PotentialUser-70, Try these steps: 1. Run `guix shell python python:tk' 2. Run `python3'. 3. In the Python REPL, run `import tkinter'.
<podiki>sarg: do submit a patch if there isn't one already for pipewire update
<ieure>PotentialUser-70, These steps work for me, and indicate that the correct packages are set up and usable. If that works for you, the issue is likely in your application.
<peanuts>"[PATCH 0/2] Add support for unlocking root device via a key file"
<ieure>PotentialUser-70, Note that `guix shell' makes a subshell with a more restricted environment. You'll need to `exit' or hit Control-d to get back to your normal setup.
<podiki>graywolf: ah. several people do use encrypted drives with guix, would be good if one of them could review (I use a keyfile on another distro but don't think I could provide a useful review here); perhaps ping guix-devel since this would be of interest to several people
<vivien>lilyp, it appears gnome-shell bundles a few things, including gnome-volume-control and Shell External Windows
<peanuts>"[PATCH] gnu: pipewire: Update to 1.0.0."
<Franciman>sarg: you are a true hero, siocan
<podiki>thanks; will see if QA gets to building it soon and take a look
<PotentialUser-70>ieure, thank you, ot works in shell and I am trying to figure out why it doesnt work if I start the project in this shell
<sarg>PotentialUser-70 tk is in a separate output: guix shell python:tk -- python3 -m tkinter
<lilyp>vivien you mean "extension", or…?
<cwebber>heck yeah to dthompson getting Hoot 0.2.0 (Guile Scheme -> WASM) in Guix *already*!
<cwebber>$ guix shell guile-next guile-hoot
<peanuts>"Guile Hoot v0.2.0 released! -- Spritely Institute"
<cwebber>exciting times :D
<ieure>cwebber, Super exciting! Really appreciate y'all's work on this stuff.
<Kolev>cwebber, thank you so much for getting Scheme into the browser! I wonder how JavaScript Web Labels can work with WASM, though.
<vagrantc>ACTION hoots
<Kolev>How to mark WASM as free, like LibreJS?
<cwebber>wingo_, dthompson, and robin did all the hard work tho I should say :)
<ieure>Kolev, I don't think that's relevant, WASM is a binary format, it's not source or source-equivalent. Does Guix put license info in the ELF header of its binaries or something?
<dthompson>no one really uses web labels anyway and often it's not possible to satisfy librejs even when you want to
<dthompson>it was briefly my job to try to get fsf hosted web apps librejs compliant.
<dthompson>and ieure is completely right. wasm is a binary format. though javascript has been essentially a binary format in source form for many years now.
<Kolev>ieure, LibreJS points to the "binary" (minified JS) and the source, and the license. It shouldn't be hard to point to the binary.
<Kolev>dthompson, yeah, I think the real solution is to not run any scripts at all and to install a web extension for every site that needs scripts.
<Kolev>Let Guix handle website scripts.
<ieure>Kolev, That's more or less what I've been doing for the last ~4 years. Using NoJS in default-deny mode.
<Kolev>The web is so hard to use.
<ieure>You really get to see how poorly sites are built if you go that route. Dozens of third-party scripts all over the place, and everyone has 2-3 domains (,, etc) because 20+ years ago, someone decided browsers couldn't have more than 4 concurrent connections to the same server. Or whatever the limit is.
<ieure>At no point has someone deigned to lift this limit, so everyone just pays 2x+ in domain fees to host stuff. Awesome.
<Kolev>Maybe Hoot can output a WebExtension of the WASM? That would be cool.
<Kolev>This way, a site's script can be easily managed by Guix.
<robin>options for license info might include http link headers, web labels listing wasm files as "minified" versions, or a custom section in the wasm file with license/source info. idk of anyone using those yet though
<Kolev>dthompson, sorry you're soured on Web Labels. But I do know what LibreJS can do to a person; it's rough, dealing with LibreJS.
<zilti>ieure: That connection limit has stopped mattering a loong time ago
<ieure>zilti, Not sure what you mean. It still exists and is an issue even in modern browsers, which is why everyone registers extra domains. It's not an issue with HTTP/2, but lots of stuff isn't on HTTP/2 or has backwards compatibility which requires the domain sharding.
<ieure>It stopped being *a good idea* a long time ago. But it still exists, therefore, it matters.
<ieure>Though perhaps if webdevs actually cared about performance, they be shoving 15mb of iffy JavaScript into everyone's browser.
<Kolev>If I ever need client-side code on a website, I'm going to release it as a WebExtension.
<apteryx>still no dice about e1000e connection establishment bug on x200; I've left a comment here:
<peanuts>"talk -"
<apteryx>weird, 'guix weather qemu-minimal' says 0 substitutes available on a machine
<apteryx>freshly guix pulled
<apteryx>CI disagrees:
<peanuts>"Search results"
<nathan-lt>I want to use the `power-profiles-daemon` package, but I'm not sure of how to do so. Instructions for other distros involve starting a systemd service, but I don't see a corresponding service in guix. Do I have to write a service to be able to use the package?
<nathan-lt>Or does merely installing it on guix configure everything necessary?
<ieure>nathan-lt, You have to reconfigure the system to change what services run.
<nathan-lt>ieure: I thought so, but I don't see a service for the package. For example, there is `tlp-service-type`, but I don't see a `power-profiles-daemon-service-type`.
<apteryx>is there a way to tether networking to a second Guix System via USB?
<apteryx>perhaps networkmanager knows to do that? has anyone tried/can report about it?
<ieure>apteryx, Haven't done it with Guix, but Linux supports USB Gadget Ethernet.
<peanuts>"How to Use USB Gadget Ethernet | Timesys LinuxLink"
<ieure>apteryx, I have a raspi with some stuff that exposes this, plug a cable into its MicroUSB port and the other end into a computer and it shows up as a USB Ethernet interface.
<apteryx>will read on it, thanks
<anthk_>on nm, I wish there was an easier way to set vpn plugins for it
<Kolev>apteryx, the Guix System installer recognizes USB tethering.
<apteryx>TP 150N USB wifi dongle also stopped working on that X200 machine:
<peanuts>"ath: phy82: Failed to wakeup in 500us ? Issue #72 ? qca/open-ath9k-htc-firmware ? GitHub"
<apteryx>so USB tether will be the last option!
<Kolev>Hm. I don't see any WebExtensions packaged for Guix, such as Adblock or NoScript.
<ieure>Kolev, There are uBlock Origin packages for IceCat and Chromium.
<graywolf>When trying to apply a patch, I am getting "File test/test_torrents/v2_empty_file.torrent: git binary diffs are not supported. ". Is there a common way around that?
<ieure>Kolev, I agree that Guix packages for more browser extensions, and build-systems to help ease that, would be a very good thing.
<graywolf>(in the (package (source (patches ...)))
<Kolev>ieure, thanks. I was looking for a packaged WebExtension to inspect.
<vivien>lilyp, in gnome-shell-44.5.tar.xz, I see 4 subprojects: extensions-app, extensions-tool, gvc (“libgnome-volume-control”), and shew (“Shell External Windows”)
<Kolev>ieure, a shame WebExtensions are attached to a specific browser. I thought they were browser-agnostic now.
<ieure>Kolev, I don't know too much about that. I do know that Chrome is nerfing the ability of extensions to block ads, so perhaps there's a WebExtension for Firefox-based browsers, and one that does what it can with Manifest V3 for Chrome?
<ieure>*Chrome-based browsers
<Kolev>I say "Mozilla" and "Chromium."
<PotentialUser-77>i remember hearing that epiphany can use firefox extensions
<PotentialUser-77>and they have different rendering engines
<attila_lendvai>i guix pulled, and my desktop fails to come up. i think the root cause is that NetworkManager doesn't start. am i alone with this?
<graywolf>How serious is the "run guix style" part of contributing manual? The output it produces... is weird, and seems to introcduce quite a lot of noise into the patch I would sent.
<vivien>Those “this unmaintained project must be used as a git submodule” projects are getting really annoying now
<civodul>graywolf: ‘guix style’ often does a good job, but sometimes not; so use your judgement i guess :-)
<Kolev>What is it?
<civodul>also, if you find cases where you think it misbehaves, you can report them as a bug
<Kolev>What is guix style?
<graywolf>Right, so I will just ignore it this time. So I can move to step 5 out of 16 :)
<civodul>makes sense :-)
<civodul>Kolev: a tool for code formatting:
<peanuts>"Invoking guix style (GNU Guix Reference Manual)"
<graywolf>civodul: Since you are here, would you happen to know how to correctly use search-patch with local-file inside snippet?
<peanuts>"debian Pastezone"
<graywolf>This gives warning :/
<vivien>Is there a standard pattern for using source code libraries in a package? Something like, in a source snippet, unpack the source of that library in this subdirectory
<vivien>That way we can at least patch the dependency
<ieure>vivien, I'm by no means an expert, but I'd think you would want one package for the stuff in the submodule (which ships the source code, not a binary), then use that in the inputs of the things that needs it.
<ieure>Someone else probably knows better, that's just my hot take.
<vivien>If you use it as an input, it will be compiled separately
<podiki>vivien: yes, we do that, sometimes with a package that is just source (no build), a bare origin (just to grab the source), or using package-source from an input
<vivien>Do you have an example?
<podiki>here is one for an input that is just an origin to get the vulkan headers (needed because it is a different version required):
<peanuts>"guix.git - GNU Guix and GNU Guix System"
<podiki>and used to unbundle the included vulkan
<podiki>rather than fetching a submodule (which we can't do)
<podiki>another would be ogre, which uses package-source in its native-inputs to get the source from a package (unpacked in a phase)
<vivien>Thank you :)
<podiki>there's plenty of that around, as you've observed submodules have spread and make proper packaging needlessly more difficult
<Kolev>Does emacs-guix work still?
<ieure>vivien, That's why I said the package for the submodule should ship the source. A package build doesn't imply source compilation -- you can use copy-build-system to produce a build that just includes files.
<ieure>Kolev, Yes, but it's kind of buggy. It breaks tab completion in shell-mode buffers.
<podiki>ieure: we do have some packages that are "source only" but more typical I think is to use e.g. package-source to get the source used in another package (which can be built on its own as usual)
<ieure>podiki, Ah, interesting. I didn't know that, thanks for the info.
<podiki>I think the comment was that including something as an input means it gets built first, which is true as far as I can tell
<podiki>so if you need the source, and not maybe just headers that are part of the output, then you do these other things
<podiki>I think a bunch of go packages aren't built but are just source...? (I remember having comments about "source only" packages in our go land, where build is skipped)
<ieure>podiki, The word was specifically "compiled." Not all package builds compile.
<ieure>I agree that it builds the package; I disagree that build implies compile.
<podiki>as you say, copy-build-system is the prime example, most other build systems you will do some compiling (or equivalent) unless removed
<podiki>since we are a source-based distro, we always have source, just usually via package-source rather than source-only packages
<podiki>(as opposed to other distros that do -dev packages or other things, separating out headers from built libraries, etc.)
<Kolev>How do I use Guix Shell from Guile?
<ieure>Kolev, That's a somewhat odd question. Can you explain what you're trying to do? There might be a better way to go about it.
<Kolev>ieure, do in Scheme.
<ieure>Kolev, Interesting. Maybe look at the code for the guix shell command?
<PotentialUser-77>the shell commands are in /guix/scripts/
<lechner>Kolev / i'd use system*
<Kolev>lechner, how does `system` differ from `system*`?
<Kolev>lechner, so (system "guix shell ...")?
<lechner>Kolev / it's more or less a straight fork & exec and bypasses shell expansion. It basically means no quotes but also no piping
<graywolf>Kolev: (system "echo foo") (system* "echo" "foo")
<lechner>system is stringy
<graywolf>If you do not need shell features, use system*
<lechner>you can get the environment in something like (let ((home (getenv("HOME")))) ... )
<graywolf>How long it usually takes to receive the issue number for a cover letter?
<ieure>graywolf, A few minutes.
<graywolf>So if it is 20 already, I assume I did something wrong.
<graywolf>Will try to send it again.
<ieure>Check your spam folder too.
<graywolf>Hm, the log looks fine (status=sent (250 OK id=1r8qKy-0004dw-TD)), but I see nothing in the other direction
<lechner>graywolf / did you send to or
<lechner>also, have you submitted before?
<graywolf>Not sure if from the new email address
<lechner>i would not send it again. it can take up to three hours
<graywolf>welp, it that case rest of the series will go out tomorrow :)
<lechner>what's the topic please?
<graywolf>libtorrent-rasterbar is pretty much unbuildable currently, which breaks qbittorrent and deluge
<graywolf>So I fixed that and upgraded to new versions
<graywolf>Or did you meant the email subject?
<peanuts>"#67552 - libtorrent-rasterbar tests fail [with patch] - GNU bug report logs"
<lechner>ArneBab beat you?
<graywolf>Yeah, that just disables the tests, not exactly elegant solution... Albeit a quick one.
<lechner>the right thing is to close your bug and amend his
<lechner>so you can send now
<Kolev>lechner, I'm missing some pieces.
<lechner>graywolf / also, i agree with your assessment
<lechner>you are T. Volf?
<graywolf>Yeah :)
<lechner>thanks for that ibus help some time ago!
<lechner>it was on the list
<graywolf>Yeah no problem :) I just recalled that I ran into the exact same issue when I was setting my machine
<graywolf>It really should be documentend (somewhere?)
<graywolf>Or even better, it should just work
<lechner>it was a real issue for me. we speak mandarin at home and my son was unable to do his homework
<Kolev>lechner, thanks for speaking another language at home.
<lechner>it didn't work under GNOME, however, and only under EXWM (my machine). Do you use GNOME?
<graywolf>Oh I see, that sound unpleasant. For me, I am just learning Japanese as a hobby, so it was just an annoyience
<graywolf>No, I am running i3, directly from login shell (via my custom startx)
<lechner>my wife is from china as are most of our friends, but i'm from germany and live in california
<graywolf>I never really saw any need for full desktop manager
<lechner>yeah, GNOME ships ibus i think and i can't get it working either way
<lechner>Kolev / what do you speak at home?
<Kolev>lechner, English.
<lechner>Kolev / perhaps the '(list (@@ ... should be inside double quotes?
<graywolf>I find gnome... bit heavy for my taste :) For regular floating wm I typically reach for xfce
<graywolf>lechner: How do you like the US?
<lechner>not sure it's worse than any other place right now. been here for thirty-one years
<Kolev>lechner, what about ^DISPLAY$ etc.?
<graywolf>Those as well (likely). In scheme strings are just "", you cannot pick between " and '
<lechner>Kolev / those looks fine to me but i am surprised that your .desktop file is in a multi-arch folder
<graywolf>Oh that is a long time
<graywolf>Ok, I just learned that '^DBUS_' is a valid symbol syntax: $1 = ^DBUS_'
<Kolev>lechner, it's in x86_64-linux-gnu.
<lechner>graywolf / the '(list (@@ is an argument to -e so it should read "-e" "'(list (@@ (gnu packages commencement) gcc)"
<lechner>graywolf / in Guile or in the shell?
<graywolf>In guile
<lechner>what does it mean?
<lechner>Kolev / actually, it should read "^DISPLAY$" etc
<graywolf>No I am just reading the paste Kolev send (; I expected '^DBUS_' to be a syntax error
<lechner>no symbols, just strings
<graywolf>but it is valid (albeit wront in this context)
<lechner>Kolev / try the guile REPL (preferrably with readline)
<graywolf>Kolev: > In scheme strings are just "", you cannot pick between " and '
<lechner>graywolf / i think you and i are on the same page
<lechner>Kolev / also the options with = you need to string-append with a = in between
<lechner>like --share etc
<graywolf>Aaand I on my way to bed, it is getting late in Europe. Nice chatting with you, see you around :)
<Kolev>lechner, even uglier.
<lechner>graywolf / bye
<lechner>Kolev / for the constants, you can just use one big string, i.e. "--preserve=^DISPLAY$"
<jaeme>What are good USB wifi adapters for laptops with nonfree wifi cards?
<jaeme>Preferably in the less than or equal to $30 price range
<lechner>for that kind of money i would take out the old one. usb ports are valuable, and can break
<lechner> is your friend
<lechner>but blobless is tough with 5G.
<lechner>some folks say that ath9k is the juice
<lechner>this is 450 chipset,
<peanuts>"Technoethical N450 Dual Band Wi-Fi mPCIe Card for GNU/Linux - Technoethical"
<lechner>you can type ar9380 into aliexpress
<apteryx>jaeme: I've used the TP 150N USB wifi adapter just fine for a long time, but currently it doesn't work well on my x200, not sure why (it looks like this:
<peanuts>"ath: phy82: Failed to wakeup in 500us ? Issue #72 ? qca/open-ath9k-htc-firmware ? GitHub"
<jaeme>I see, I have a x220i laying around still with a blobbed intel chip, I think I'll just replace the wifi card inside.
<jaeme>Seems much cheaper
<jaeme>I'm just waiting around for rust-team's latest rust compiler 1.73 to merge
<Shaq>I'd like to have a non-tech question here:
<Shaq>Guix release under AGPL, does it means all programs I run on it under obligation of AGPL?
<mange>I don't think Guix is under the AGPL, is it? The website is (hence the note in the footer), but COPYING in the Guix repository itself is GPLv3, and the headers on all the files say GPLv3 or later.
<Shaq>I think it's under AGPL:
<peanuts>"COPYING\website - guix/guix-artwork.git - Artwork and web site of Guix"
<mange>That's guix-artwork.git, not guix.git. That's the website sources and some other visual stuff.
<Shaq>Thanks for point out my mistake, peanuts
<peanuts>Shaq: Hi, for comments please contact my maintainers at
<mange>At any rate, the follow-up question is presumably "is all software installed through Guix under the GPL?" I'm pretty sure the answer to this is no. I'm not familiar with the reasons, but if we look at the GPL-Incompatible Free Software Licenses list on we can find software packaged in Guix that uses those licenses.
<Shaq>Thanks a lot.
<Shaq>If it follow the common linux license(GPL with excpetions(e.g.  link to system library)). I am quite happy to make guix as my production server :)
<mange>I think it's more that packaged software doesn't really link to Guix. I imagine tooling built on top of Guix (i.e. importing the Guile modules that Guix defines) would have to be under the GPL, but the software that Guix packages isn't really aware of Guix at all.
<Shaq>Got it, thanks mange
<adanska>Hi Guix!
<Kolev>adanska: Hi! 😀
<bumble>hello hola and konnichiwa (japanses input broken on this guix machine atm)
<adanska>hi Kolev!
<lilyp>vivien: I don't see that as a problem: they're part of gnome-shell and not maintained elsewhere
<vivien>gnome-volume-control claims to also be used in gnome-settings-daemon and gnome-control-center
<ArneBab>lechner: yes, but I prefer the better fix.
<ArneBab>(= not mine ☺)
<Guest14>How would I use nouveau with video encoding acceleration?  For example using mpv to watch a h264 encoded video?
<efraim>Guest14: you could probably test it out with --vo=gpu
<zilti>Does someone have an example of `greetd-wlgreet-session`? Whatever I try, I can't figure out what to set as `command`.
<adamnr>Hi Hurd
<adamnr>Looks like the IRC logs for guix and hurd have stopped
<isaneran>is there an irc for hurd`
<janneke>isaneran: #hurd?
<isaneran>janneke: thanks!
<PotentialUser-29>Hello all! I have a question that I haven't found the answer to in the manual: what is the difference between editing channels.scm manually and extending home-channels-service-type? Is one approach better than the other?
<isaneran>if you're using guix home to declare your home it might make sense to use the home-channels-service-type
<isaneran>though at the same time, I am not entirely sure if you'd be able to install all the packages in your home if you configure it this way
<isaneran>because if you're importing modules from a channel that hasn't yet been downloaded, the home file might fail to run
<isaneran>civodul: am I wrong that this could happen?
<roptat>hi guix!
<roptat>you're right, but you could do it in three steps: home reconfigure to set up your channels, pull to get the new channels, home reconfigure again with the complete configuration
<isaneran>ah so you make two home configurations
<isaneran>one that just sets up the channels
<isaneran>and one that sets up the channels and packages and configurations etc
<isaneran>they could load the part for setting up channels in a shared file too so
<isaneran>Hmm I might do this
<isaneran>then on a new computer I can just do `guix home reconfigure bootstrap-home.scm && guix home reconfigure my-home.scm`
<PotentialUser-29>Thanks for the answers! So from what I understand it's more complicated to use the service, so what would be the benefit of doing that? Having everything in one place maybe?
<Guest14>efraim: Nope, not working
<Guest14>i also ran guix gc and reconfigured home again.  it needed to download everything again.  shouldn't guix home link it to prevent guix gc from removing the store items?
<Guest14>3D acceleration isn't working either
<bienjensu>Could you webpaste your system & home configs somewhere? (e.g. <>)
<peanuts>"OK, turing."
<old>anybody had problem with geiser recently? It just hang for ever after every prompt
<isaneran>in the repl?
<isaneran>which scheme impl? guile?
<isaneran>no matter what the input?
<old>Just evaluating something like `1' or `a'
<isaneran>works here
<old>it never print the result
<isaneran>and I updated everything like 1 hour ago
<old>must be something in my emacs config then
<old>here I go again bisecting emacs
<isaneran>maybe you have old bytecode or native code cached
<isaneran>or maybe *I* do and I happily have working geiser, for now
<Guest14>Just to be sure, 3D accelerations should work out of the box with linux libre and nouveau?
<old>isaneran: I've deleted everything under .emacs.d to ensure that's not the case. I'm commenting everything in my configuration now to determine the culprit
<isaneran>Guest14: all of them?
<isaneran>not sure if it would work under libre because of firmware stuff
<isaneran>old: does it work with a minimalist enviroment where you install nothing but geiser guile and geiser-guile and emacs?
<isaneran>e.g. in a guix shell
<old>isaneran: That's what I'm trying now with guix-home container :-)
<isaneran>well, sounds like I can't be much more than a rubber duck for you :D
<old>happy to see I'm not too much crazy in my investigation workflow :p
<old>hmm okay so something in my Guile configuration is messing up with geiser
<old>I will try in a pure environment
<isaneran>makes me curious about what you're putting in your guile config
<old>okay so with shell --pure --container emacs emacs-geiser that's fine
<bienjensu>Guest14: It's difficult to help without a more precise definition of the problem, access to your system and home config files, and knowledge of what hardware you have.
<old>something something in my config is messing with geiser
<isaneran>old: watch it be emms or something for no reason at all
<bienjensu>old: If emacs is genuinely hanging, you could try using the profiler to figure out where the problem area is.
<isaneran>old: or why not the color theme you're using
<old>isaneran: I doubt that modus operandi is the problem here :p
<isaneran>speaking of colour themes I've set up so I get different ef theme based on the season, and since today it is 1st of december I now have the winter theme B)
<isaneran>old: probably not, though I've had weird issues where after loading a theme my org agenda suddenly stops working etc
<isaneran>I guess configuration being code has both upsides and downsides
<old>right. you only know the 1st of december that something break :-)
<old>okay seems that the culprit is within my setup-lisp-family hook
<isaneran>ef-themes and modus themes are from the same guy and he seems to know what he's doing so
<old>which is called for scheme-mode, emacs-lisp-mode and geiser-repl-mode
<isaneran>well at least the issue was in something related
<old>something logical
<isaneran>was it bienjensu's proposal that helped you locate it?
<isaneran>or did you just divide and conquer / binary search your config
<old>isaneran: good old binary search
<old>never failed me
<old>(paredit-mode 1) is the culprit wtf
<isaneran>oh shit
<isaneran>and I just re-installed paredit for the first time in a long time
<isaneran>good thing I didn't activate it yet B)
<old>but why
<old>it was working fine before!!
<isaneran>e m a c c c c c c c c c s
<old>paredit in a geiser-repl window was working fine fine fine like a week ago
<bienjensu>Try a minimal container with paredit and geiser?
<isaneran>also, isn't paredit like, basically "finished"?
<old>bienjensu: done and I can reproduce
<isaneran>does it even get updated anymore?
<old>guix shell --pure --container bash emacs emacs-geiser emacs-geiser-guile emacs-paredit
<bienjensu>Which commit of guix are you on?
<bienjensu>`guix describe`
<old>then start emacs and M-x run-geiser then enable paredit in the REPL and voila hang for every
<old>eneration 42 Nov 30 2023 16:51:39 (current) guix 2b782f6 repository URL: branch: master commit: 2b782f67266b42bb40015bd23ce2443be2f9b01f
<peanuts>"guix.git - GNU Guix and GNU Guix System"
<isaneran>old: M-x run-guile no?
<isaneran>ah there is both
<old>isaneran: righ right
<isaneran>I remember like 9 years ago run geiser screwed up but run-guile didn't
<isaneran>so I guess I've stuck to that
<isaneran>Please let me not gc away the guile debug 3.0.9 every time lol
<old>I'll check if that's the update of emacs-geiser in september that did that
<old>so did anyone reproduced on their end?
<isaneran>still waiting for downloads
<isaneran>ok I have the shell
<isaneran>display or term aren't set so I can't start emaccc
<old>hmm not the new version of geiser the problem
<old>isaneran: I ran the container in a tmux panel
<old>then I do: TERM=tmux emacs
<old>but you can also pass the TEMR environment to the contianer
<old>guix shell --preserve=TERM
<isaneran>I thought container didn't have access to my config file
<zilti>Does someone have an example of `greetd-wlgreet-session`? Whatever I try, I can't figure out what to set as `command`.
<isaneran>it's complaining that it can't find pdf tools etc
<isaneran>either way it starts when I just set it to a random term
<isaneran>my terminal is foot and apparently that doesn't work
<zilti>I always just get "invalid G-expression entry"
<old>isaneran: right. the container will only have access to directory tree under which you started it. For the terminal to work, just pass the TERM variable or set it to something like xterm-256color
<isaneran>ah ok so I need to start it somewhere else
<old>well you do not need your config file for this
<isaneran>btw you didn't include paredit in the command haha
<isaneran>I can just eval the requires
<old>realy the following command ought to work: guix shell --pure --preserve=TERM --container bash emacs emacs-geiser emacs-geiser-guile emacs-paredit -- emacs .
<old>then in emacs you do:
<old>M-x run-guile
<old>In the Geiser repl:
<old>M-: (paredit-mode 1)
<isaneran>with slightly different steps
<isaneran>though emacs itself isn't frozen
<isaneran>just the prompt
<isaneran>I guess that's what you said at the beginning
<old>emacs is fine, but the geiser prompt is dead
<isaneran>for me: in a new directory, ran guix shell --pure --container bash emacs emacs-geiser emacs-geiser-guile emacs-paredit
<isaneran>then I ran TERM=screen emacs
<isaneran>then I ran M-x geiser-guile
<isaneran>M-x paredit-mode
<isaneran>Ah, geiser-guile doesn't become available after trying to run one of the deprecated command shaha
<isaneran>either way, M-x geiser and switch to the repl
<isaneran>so it's not just your machine
<old>not sure if this is a guix bug then. I should probably ask if this can be reproduced on other system on #emacs
<isaneran>good idea
<isaneran>I suspect it's an issue with geiser
<old>It's sad that I have to disable paredit in geiser tho for now
<old>it was neat
<old>isaneran: anyway, thanks for you help in the investigation :-)
<isaneran>is geiser-repl-autoeval-mode active old?
<isaneran>the docstring for it says
<isaneran>This mode may cause issues with structural editing modes such as paredit.
<isaneran>there is also geiser-squarify that does some paredit specific magic
<isaneran>but that's for toggling between () and []
<isaneran>seems unlikely to be the cause
<old>Its value is nil
<isaneran>ok so when I am not using paredit enter is bound to geiser-repl-maybe-send, when I am using paredit, it is bound to paredit-RET
<isaneran>that's probably the issue
<isaneran>if I M-x geiser-repl-maybe-send I get the result of the evaluation
<isaneran>so it's probably just that paredit inserts a newline
<isaneran>but geiser is waiting for the commmand
<isaneran>old: I think you could fix it by adding a hook to paredit mode that checks if it's in a geiser repl and rebind enter key
<old>isaneran: aaah right
<old>how I can do that I can't rembmer
<old>changing a binding for the buffer
<isaneran>let me hack something together then
<old>isaneran: thanks
<peanuts>"debian Pastezone"
<isaneran>this works but, looks like it sends the input even if you wanna put a newline
<isaneran>ah I guess that's because the expression is always balanced
<isaneran>probably works better if you set geiser-repl-send-on-return-p to nil
<isaneran>yep then it works perfectly
<old>let me try this
<old>ah you're doing this in a paredit-mode-hook
<old>I was doing this in my list-setup-family hook
<isaneran>cuz when you run M-x paredit-mode it sets enter
<isaneran>binds the enter key*
<isaneran>and we don't want to do this for a normal scheme buffer anyway
<old>thank you so much!
<old>you made my day
<isaneran>yay \o/
<isaneran>no problem
<isaneran>happy I was able to be more than a rubber duck after all
<old>I'm wondering, I'm a the only one using paredit in geiser?
<isaneran>probably not
<old>otherwise, seems to me other would have got that issue, yet this problem is not mentionned anywhere in geiser documentation
<isaneran>problem is, I'm not sure who's bug this is haha
<old>none I guess. It's a configuration bug
<isaneran>yeah, maybe it should be documented in geiser manual though
<old>would not mind ^^
<old>thanks again! I have to go now
<old>good day to you !!
<isaneran>but, I was quite saved by the good documentation by using the help system on the maybe-send command and the variable you had to change the value of
<isaneran>so the code itself seems fairly well documented
<isaneran>yeah I also have to go!
<isaneran>have a nice day!
<attila_lendvai>is there a string backquote facility in the guix codebase? like ` but not for lists but strings? my ultimate goal: use variables in my nftables config for the lan address and similar.
<attila_lendvai>format is unmanagable with positional arguments, and i don't want to use a large (string-append ...)
<gabber>your nftables config as in some file on your fs or as part of your guix config?
<attila_lendvai>gabber, as part of my guix config
<gabber>maybe it's easiest to write a small helper function which does the (format) for you?
<attila_lendvai>gabber, the issue remains the same: large literal string with scattered points where i want to insert stuff from the scheme universe
<gabber>i'm not sure i completely understand the problem and your intended solution. would you mind giving an example?
<gabber>if it works for lists you could work on those and only in a last step convert them to string...
<attila_lendvai>a fictious solution: (plain-file "nftables.conf" #`"large literal nftable string #,%a-scheme-variable some more literal string")
<mirai_>something like srfi-166 perhaps?
<sneek>Welcome back mirai_, you have 3 messages!
<sneek>mirai_, lechner says: / Hi, how should I deprecate krb5-configuration (which I renamed) please? Using define-deprecated gives an error. Thanks!
<sneek>mirai_, lechner says: / nvm, i referred to the configuration in record syntax <> instead
<sneek>mirai_, lechner says: / actually, deprecating records does not work gracefully. it errors out because the deprecated record does not exist. ("error: krb5-configuration: unbound variable") Is there a solution?
<peanuts>"SRFI 166: Monadic Formatting"
<attila_lendvai>mirai_, yeah, SRFI 166 looks interesting. first time i looked, thanks!
<peanuts>"SRFI 166: Monadic Formatting"
<jackhill>hmmm I notice evince (both master and gnome-team) with sway with a scaling factor 2 doesn't fill the screen when in presentation mode. I wonder if that's our problem or gtk/sway's
<jackhill>I've asked if anyone else has noticed this in #sway
<vivien>lilyp, to be sure, gnome-shell 45.1 with mutter 45.1 also fails :(
<bienjensu>Is gsettings removed from glib purposefully?
<nckx>bienjensu: It's not removed.
<nckx>The main output is for the primary (library) use case.
<bienjensu>oh! ty
<bienjensu>`guix locate` didn't, but probably should, find it
<nckx>I didn't know gsettings was a glib® product TBH. So thank you.
<bienjensu>I was tipped off by the bash completion file in the main output.
<nckx>I'm not a ‘guix locate’ user, but doesn't it only search installed packages/profiles/something else?
<nckx>Yep, that's all it returns for me as well.
<bienjensu>Yeah, even with glib:bin in the store it doesn't find gsettings.
<lilyp>vivien: the problem lies with the test inputs or whatever's shadowing them
<nckx>guix locate -um store && guix locate gsettings # but yes, this takes ages & isn't what they call in the business the great UX
<nckx>bienjensu: ^^
<bienjensu>I wasn't aware locate only traversed the local store. I'm usually hunting for a utility that is bundled with a package I know is 'being used' (as a dep e.g.) but hasn't been propagated into my profile, which works well.
<bienjensu>nckx: ah
<nckx>It does only traverse the local store, but if something is being used, it should be in the store, no?
<bienjensu>nckx: As in, things might depend on alsa-util, for instance, but alsa-util (which has aplay) might not itself be propagated into the profile.
<bienjensu>so its /bin isn't in your path
<nckx>Hence the ‘-m store’, so it searches the local store, not just a few profiles.
<nckx>Unless you mean it still won't be found then (I'm still updating my cache).
<nckx>Honestly at this speed it's arguably better to ls /gnu/store/*-*-*/bin/gsettings unless you like to locate things, like, a lot.
<bienjensu>Maybe I hadn't rebuilt my cache.
<bienjensu>only 30k packages in my store, maybe mine will build before yours :)
<nckx>Oh dear. I'm ‘indexing 28,562 packages’ but I'm on an unoptimised file system and have zero NVMEs or whatever, so maybe you'll still win.
<vivien>lilyp, by using the incredible trick of adding #:test-options '(list "--verbose") to the arguments, I noticed the tests choke when trying to run the gsd-color dbus mockup, whose definition is in mutter’s output
<vivien>(funny thing, it’s the exact same place where gnome-shell 45.1 / mutter 45.1 fails)
<vivien>So, my 2 cents is something something gnome-settings-daemon something
<bienjensu>nckx: locate still doesn't find it
<nckx>Here it's still indexing.
<nckx>And boo.
<nckx>ACTION away.
<adamnr>Hi Guix
<adamnr>guix & hurd IRC channel logs have stopped logging
<Franciman>has hurd first class support in guix?
<peanuts>"IRC channel logs"
<Franciman>ah no, sorry. Has*
<avp>sneek: Later tell civodul: Hello! I prepared a fix for Kubo build:
<peanuts>"#67570 - [PATCH] gnu: kubo: Fix build. - GNU bug report logs"
<lilyp>vivien, you may not like it, but 42.2 is already the latest version of gnome-settings-daemon
<lilyp>at least to guix refresh it is