IRC channel logs

<Guest27>Now will say what the error is

<Guest27>Error is unbound variable

<Guest27>Use-service-modiles: unbound variable

<ieure>Guest27, It looks to me like if you have a (file-system ) in your system configuration which uses bcachefs, the initrd stuff will include the correct module for you automatically.

<Guest27>Hint did u forget a use modules form?

<ieure>Okay, really have to go now, later.

<Guest27>Ok

<Guest27>Btw I dont have a filesystem which uses bcache yet

<Guest27>I wanted to create one

<Guest27>Ok

<Guest27>For whoever is reading this and can help.me: I didnt touch the use-service-module line but the error is about thay line

<Guest27>Idk why

<Guest27>Ok ieure the problem was that I needed this line (use-modules (gnu) (gnu packages filesystems)) and I erroneously thiught that I ciuld remove the first gnu because it was repeated twice haha

<Guest27>Thanks for your availability

<YAR_Oracool>Hmmm... I have to sand box a few programs

<bjc>can someone remind me where guix keeps the public/private keypair for offloading?

<mra>howdy guix! o/

<mra>still running into some issues with guix pull. seems like it's some kind of issue with guix substitute? running guix pull --no-substitutes works, but allowing substitutes causes it to fail with an error. i don't really want to solve this problem by bootstrapping everything...

<mra>i posted it the other day, but the error i'm getting is here: https://paste.rs/eynWz.txt

<mra>someone proposed that it might be a channels issue, but i tried manually specifying the channel to use and the problem persists

<bjc>nm, i found it: /etc/guix/offload

<bjc>mra: just guessing, but it sounds like something in a gexp is calling gcrypt which is getting a bad hash or something

<bjc>it /etc/guix/acl ok?

<bjc>are you adding any custom substitute servers?

<mra>bjc: it contains two public key entries. no custom substitutes that i know of? if i somehow added some and then forgot, where would they be?

<bjc>dunno. i *think* it's just the acl file that gets consulted, but i'm not sure

<mra>oh, hm... i wonder if this is an issue with berlin...

<mra>i ran "guix weather coreutils" and got this output: https://paste.rs/6wVFG.txt

<bjc>i wouldn't think so. it sounds like gcrypt is trying to work with an ipappropriately sized block for whatever reason

<mra>yeah, i just can't figure out why that would be happening

<bjc>oh, hrm. you're running this on nix out of default profile

<bjc>guess the first guix pull didn't work for you

<bjc>or, rather, this *is* the first attempt at a pull

<mra>bjc: yeah, running on nix. this is the first attempts at a pull

<mra>i need to build some swift software, and nixpkgs doesn't have a new enough version of swift packaged, but guix does, so i was going to use guix to do it

<bjc>if you can get the build deps, you might be able to work around it by doing a dev setup

<bjc>./pre-inst-env guix pull

<bjc>assuming whatever the problem is has to do with the guix version you're using

<mra>oh, that's an idea. i've done that locally before when working on zfs stuff

<mra>getting the deps on nix is a bit of a pain, but it's possible

<bjc>up to you. whatever is happening is very strange imho, so it's a shot in the dark

<bjc>i think i've built guix and gcc about a dozen times each over the last week trying to get hurd stuff up =)

<mra>bjc: ooh, hurd stuff! that's exciting! are you trying to get it running on a machine of yours?

<bjc>no, just in a vm

<bjc>i know people have done it on real hardware with debian, not sure about guix

<mra>what the hell... just built guix locally and ran ./pre-inst-env guix pull. same issue

<bjc>bad gcrypt? i assume it's using whatever is packaged in nix, not guix

<bjc>guile-gcrypt more likely

<bjc>if you could do the initial pull without substitutes, you'd get up-to-date guix on its own profile with guix-controlled packages, so it should be the same at that point

<bjc>but i'm just guessing, again, and that'll take a long time to do, most likely

<mra>yeah, i'm running on a 14-year-old laptop right now, so i don't exactly want to do the pull without substitutes

<mra>okay, i tried the same thing but fetching my packages from a newer version of nixpkgs. i'm getting a weird new error? "error: gcry_md_hash_buffer: Function not implemented"

<bjc>nfc on that one

<bjc>more crypto stuff though, it seems

<czan>mra: Chiming in a bit late, with two ideas. Does "./pre-inst-env scripts/guix pull" also fail? (I've found this resolves some issues for me in the past, but I don't really know the details.) Alternatively, if "guix build guix" works does "guix shell --pure guix -- guix pull" also fail?

<czan>If "guix build guix" works, then I would also expect a pull with no substitutes will only need to build Guix itself locally (not the deps), which might make that option slightly more palatable. :)

<bjc>the deps in this case are coming from nix. to have guix build guix (rather than make), the deps need to come from guix

<bjc>if you don't have substitutes, that involves a lot of building, since you have to do the whole thing from almost nothing

<bjc>if you can 'guix shell -D guix', though, then you're not far off

<czan>Right, but looking at the "guix weather coreutils" output that was posted earlier, it looks like it's just a "you asked me to talk to berlin.guix.gnu.org, but ci.guix.gnu.org replied", which seems like it could be solved with just --substitute-urls="https://ci.guix.gnu.org https://bordeaux.guix.gnu.org". The error about invalid length specifier sounds a bit like a corrupted /etc/guix/acl or /etc/guix/signing-key.pub

<czan>, though.

<bjc>the /etc/guix directory should be something that can be deleted and recreated by nix activation

<bjc>i keep my nixos install with tmpfs / to ensure stuff like that

<bjc>oh man, i'm building gcc again. i knew something was up when i saw glibc float by

<mra>bjc: welp, built guix from source with the latest versions of all of the deps from nixpkgs, and I'm getting the same damn error

<Ofthemasses>Hey, I'm having a look into using greetd on my guix system. I happened to notice that tuigreet is listed on the packages website https://packages.guix.gnu.org/search/?query=tuigreet. Yet the link shows the package is missing: https://packages.guix.gnu.org/packages/tuigreet/0.9.1/. Was it removed recently?

<bjc>mra: oof

<czan>Ofthemasses: I have tuigreet in my local guix, according to "guix show tuigreet".

<bjc>i gotta imagine it's something to do with the nix package, but i don't see a way around it without a full bootstrap

<mra>bjc: it's very strange. it looks like a guix issue on the surface, but it seems like it had to be a nixos thing

<Ofthemasses>czan Yeah same here when I do:

<Ofthemasses>guix package -s tuigreet

<Ofthemasses>name: tuigreet

<Ofthemasses>version: 0.9.1

<Ofthemasses>outputs:

<Ofthemasses>+ out: everything

<Ofthemasses>systems: x86_64-linux

<bjc>the neat thing about guix is that it's self contained, as long as it's within guix anyway. so if it were a guix issue a bunch of us would see it, presumably

<czan>mra: Do you have LD_LIBRARY_PATH set, or something?

<ieure>mra, Why is your store path /nix/store?

<Ofthemasses>czan Do you happen to have your config public, seeming you are using tuigreet? I was originally going to come in here asking how to setup greetd-service-type with tuigreet

<czan>I am not using tuigreet. I just checked whether it had been removed from Guix (which was your question).

<bjc>ieure: they're running the nixpkgs version on nixos

<ieure>mra, "i ran 'guix weather coreutils' and got this output: https://paste.rs/6wVFG.txt" -> 100% seems like bad certs on berlin.

<Ofthemasses>czan Ah I see, you were meaning that it is in your package list. No problem!

<Ofthemasses>To solve the issue: "guix system: error: service 'console-font-ttyX' requires 'term-ttyX', which is not provided by any service." Do I need to define 1 all VTs seperately to solve this, or is there a better way?

<Ofthemasses>Here is my implementation:

<Ofthemasses> (service greetd-service-type

<Ofthemasses> (greetd-configuration

<Ofthemasses> (terminals

<Ofthemasses> (list

<Ofthemasses> (greetd-terminal-configuration

<ieure>Ofthemasses, See topic, use a pastebin. Do not paste into the channel.

<Ofthemasses>Here is my implementation:

<Ofthemasses> https://pastebin.com/anyn8ErL

<ieure>Ofthemasses, That, or remove the console-font-ttyX services you don't need/want.

<Ofthemasses>ieure My bad, where is the topic? I only see:

<Ofthemasses>ieure Okay! Thank you, I might remove the services I don't need. It seems cleaner than defining vts past 3

<mra>ieure: I thought it might be a Berlin issue, but running guix pull with --substitute-urls explicitly excluding Berlin doesn't change anything

<ieure>Ofthemasses, or you can use a looping construct, like `map'.

<Ofthemasses>ieure Oddly I get the following when attempting to remove it: guix/ui.scm:921:18: error: console-font-tty4: unbound variable

<Ofthemasses>Doing this with (modify-services %base-services

<ieure>Ofthemasses, you can do both.

<ieure>Modify services, then append the list of new services from map to it.

<Ofthemasses>ieure

<Ofthemasses>Modifying my services with:

<Ofthemasses>(delete console-font-tty4)

<Ofthemasses>causes the error above

<Ofthemasses>Are you saying I must delete the service and at the same time define a greetd session for that VT will solve my error?

<ieure>Ofthemasses, I believe you delete based on the service-type, console-font-tty4 is the name.

<ieure>Timing of when you delete/add services isn't relevant to your problem.

<Ofthemasses>did some digging in base.scm. It's "console-font-service-type". deleting that has worked, but will have to see if mingetty on tty3 functions with the service gone

<Ofthemasses>Thank you for the help ieure

<ieure>Ofthemasses, No problem. Here's a map solution to reduce duplication of your code: https://paste.debian.net/hidden/7b4a144f

<ieure>Ofthemasses, It doesn't solve the console-font services, that's still doable, but a bit harder than it should be IMO, because greetd-service-type doesn't support extension -- that would make it very clean.

<mra>alright, i've given up on fixing this damn guix bug. time to bootstrap on my 14-year-old laptop lmao

<mra>wish me luck

<czan>Good luck!

<test202020>hello, some body use librewolf?

<test202020>some days ago subs for him not available. today i install librewolf and have problem onwayland.

<YAR_Oracool>I'm haveing truble with guix home

<test202020>librewolf is so much bigger

<test202020>icecat looks as expected, but librewolf UI is bigger

<YAR_Oracool>wrong type to apply: #<service-type home-dotfiles 7ff612308b80> some thing wron with 30:10 which means nothing to me

<test202020>YAR_Oracool: you can share your config?

<YAR_Oracool>We can send files here?

<mra>YAR_Oracool: are you trying to post some text? if so, use a pastebin

<mra>you can't generally send files on IRC afaik

<adanska__>\nick adanska

<YAR_Oracool>test202020: https://pastebin.com/CKnFiT28

<YAR_Oracool>Any help wuld be good since I have no idea what's wrong.

<YAR_Oracool>I changed the adress from ~ to . but that just gave me a slightly different error in the sameplace

<mra>YAR_Oracool: oh, i see the error

<mra>home-dotfiles-configuration constructs what it says, a home-dotfiles-configuration. it's not a service

<YAR_Oracool>I mean. .this guy disagrees

<YAR_Oracool> https://guix.gnu.org/manual/devel/en/html_node/Essential-Home-Services.html

<YAR_Oracool>SHould I yank it out of services?

<mra>did you read the example in the docs that you sent?

<mra>the first code block above https://guix.gnu.org/manual/devel/en/html_node/Essential-Home-Services.html#index-home_002ddotfiles_002dservice_002dtype

<YAR_Oracool>Ah carb...

<mra>all good! hopefully that fixes things

<YAR_Oracool>invalid vlue for directories

<YAR_Oracool>paobebly needs a ""?

<mra>yeah, that's not a string

<mra>'(foo) is equivalent to (list 'foo), not (list "foo")

<YAR_Oracool>ok these two are solved now there's a problem with the exclude filed

<mra>what's the issue?

<YAR_Oracool>expected empty list

<YAR_Oracool>wron type argument in position 1

<mra>YAR_Oracool: can you send the whole error? use a pastebin if it's too long

<YAR_Oracool>I also changed '(stow) to 'stow because that was causing issues before this

<mra>yes, that argument should be a symbol, not a list of symbols

<YAR_Oracool> https://pastebin.com/Edv6wq5e

<YAR_Oracool>The only thing different thing from exclude in my field is adding org to the list

<mra>YAR_Oracool: oh, wait, what's that call to append doing in the code that you sent?

<YAR_Oracool>at least the exclude in the ocs

<YAR_Oracool>docs*

<mra>yeah, i don't think that the issue is with that

<YAR_Oracool>Want the latest home?

<mra>if you could

<YAR_Oracool> https://pastebin.com/gt8BJ7Yv

<YAR_Oracool>i tried 'stow and '(stow) the firt option gives me the error you saw

<YAR_Oracool>'(stow) says it's wrong and it should be 'stow

<mra>oh, list is scoped wrong

<mra>you have (append (list ...) %base-home-services)

<mra>but the home-dotfiles-service-type service that you just defined isn't contained within the list

<mra>good ol' misplaced parens

<mra>it should be 'stow, not '(stow)

<YAR_Oracool>It would help alot if plasma sevices installed kate and I didn't have to use nano

<YAR_Oracool>So I have to remove one ) from after the list and add it at the end?

<mra>remove a ) from the end of line 28, and add one to the end of line 34

<mra>then once you get this working install emacs and paredit :P

<YAR_Oracool>All this is road to emacs doom

<YAR_Oracool>It would be more convinient if we didn't have to say service x or y when we are under cervices.

<YAR_Oracool>services*

<test202020>file is 404

<mra>YAR_Oracool: if you don't like it, you can write a macro easily enough!

<mra>btw, have you got the dotfiles thing working?

<YAR_Oracool>working in what aspect?

<YAR_Oracool>test202020: dang. it expired

<mra>YAR_Oracool: well, as in you aren't having errors any more

<YAR_Oracool>atm I'm not getting syntax stuff, instead I'm getting cannot determine provenance for current system

<mra>well, that sounds like a different problem at least

<YAR_Oracool>true.. .still a problem.

<mra>what does it say? cannot determine provenance for current system?

<YAR_Oracool>ye

<mra>hm... what does guix system describe return?

<YAR_Oracool> https://pastebin.com/TTcgmzDD

<YAR_Oracool>bunch of stuff about guix and being generation one

<YAR_Oracool>kernel number channels and so on

<mra>oh, the provenance isn't the error that you're getting

<mra>the error is "Invalid preceding regular expression"

<YAR_Oracool>syntax error ha?...

<mra>can you try removing the .org thing from the excluded files in the home-dotfiles-service thingy?

<YAR_Oracool>sure.

<YAR_Oracool>yup that did it

<YAR_Oracool>guess * isn't as universal as I thought

<mra>well, it's a regexp. * means 0 or more of the preceding character

<YAR_Oracool>So /\\.org to say all org files?

<mra>oh, you probably put *.org when you wanted... uh... ".*\\.org", i think

<mra>i think you want ".*\\.org"

<YAR_Oracool>there is a git ignore there that say \\.git.ignore

<YAR_Oracool>and git says /\\.git/.*

<mra>yeah, but those aren't file extensions

<mra>.git is a folder, and .gitignore is a file

<mra>tl;dr: in regexes, . matches any character, so .* matches any string. because . matches any character, we write \\. to escape it to mean a literal .

<YAR_Oracool>the thing you said worked

<mra>so "\\.org" would match a file with the name .org

<YAR_Oracool>dunno if it did correctly but it worked

<mra>and ".*\\.org" matches any file ending in ".org'

<YAR_Oracool>yup that's what I need

<YAR_Oracool>gonna make literrate config

<mra>awesome!

<YAR_Oracool>now I have to see while the stuff in dotfile don't show up where they have to

<YAR_Oracool>packages (type: maybe-list-of-strings) <- this is mysuspect since I didn't put it in

<YAR_Oracool>ok I finnally did it

<YAR_Oracool>had to nullify source directory

<adanska>whats the proper way in guix to define and create a new supplementary group? I want to create a 'realtime' group that I can then use to give polkit realtime permissions. However, I can't just add "realtime" to the supplementary groups field of a user account or elsewhere, becuase it isnt 'provisioned' by the system configuration in some way. How am i able to do this?

<czan>I believe you can add something like (groups (append (list (user-group "realtime")) %base-groups)) to your operating-system.

<adanska>czan, thanks! I'll give that a go.

<kestrelwx>o/

<gabber>\o

<civodul>yelninei: hi! i’m looking into fixing #:disallowed-references issues on ‘core-packages-team’, FWIW

<sneek>Welcome back civodul, you have 1 message!

<sneek>civodul, ArneBab says: thank you!

<civodul>i think after that we should look into merging pending patches and upgrading libc & co.

<civodul>WDYT?

<yelninei>civodul: What about the really due perl update? (and all the breakage that comes with that?)

<civodul>yelninei: ah well, sure, if someone takes care of it, including the aftermaths :-)

<civodul>ISTR someone +/- stepped up to do that?

<civodul>yelninei: oh, and the Guile update!

<yelninei>civodul: i fixed all 3 texinfos and various cpan packages, i am bit unsure what to do with curl because the patch to make libcurl honor SSL_CERT_FILE no longer applies when updating it (and we are already 12 versions behind)

<yelninei>the new srfi-64 might break a lot as well

<civodul>yelninei: oh nice; for cURL i guess we have no choice but to update the patch

<civodul>it’s quite critical

<civodul>for srfi-64, i fixed guile-git and there’s a pending Cuirass fix as well

<civodul>i can take care of these two at least

<civodul>(the Guile-Git upgrade is pending in master)

<civodul>(i guess i’ll have to rebase)

<kestrelwx>Pulling from local checkout I get ```- guix pull: error: invalid name: `.-f14d4e8' ```.

<yelninei>civodul: I dont know curl to rebase the patch. Nixos just configures with --ca-path=/etc/ssl/certs but this will not work everywhere

<gabber>kestrelwx: you modified a local guix checkout and are doing now something along the lines of `./pre-inst-env guix pull`?

<civodul>yelninei: maybe publish a WIP pull request for the cURL update and call for help; i might take a look if nobody else does

<kestrelwx>gabber: Yes, I'm doing `$ guix pull --url=. --commit=f14d4e8 --disable-authentication -v3`.

<kestrelwx>Well, I also had the branch specified.

<gabber>kestrelwx: maybe you need to specify the full commit-id? but i have no experience with such things

<kestrelwx>I've started with the full one, so no.

<kestrelwx>Well, this isn't a regression, I tried an older Guix, so I'll figure it out.

<yelninei>civodul: it is part of the coreutils etc update pr. I also have an open issue for the update. I have just ignored the patch completely for now as I dont know what to do about it

<gabber>kestrelwx: i am not sure what the expected behavior is when you pass both a branch and a commit-id

<civodul>yelninei: which one is the coreutils upgrade PR?

<civodul>(sorry for getting lost!)

<yelninei> https://codeberg.org/guix/guix/pulls/4058

<civodul>right

<civodul>thanks

<civodul>yelninei: could you rebase it on top of current ‘core-packages-team’ and drop the WIP label if it’s ready?

<kestrelwx>gabber: `--url=$(realpath .)` worked out.

<yelninei>civodul: define ready, i have varying degrees of confidence in the individual patches. I am pretty sure that the gawk upgrade is correct, but am very unsure about coreutils and curl

<civodul>yelninei: ok; the coreutils upgrade was impossible before Gash 0.3.2 anyway, AIUI

<gabber>i seem unable to use two different sets of keys to push stuff to GitHub so i can submit some fixes upstream for the package i'm trying to mainstream to guix. GitHub won't accept me using two different accounts with the same pubkey and identifies me as the first account even when i test the connection with `ssh -T git@github.com -i path/to/new/identity-file`... any ideas what i could be missing? i use home-gpg-agent to handle my keys

<yelninei>civodul: I hope I rebased correctly as the label change really messed things up

<civodul>thanks!

<kestrelwx>When it comes to getting `SIGILL` during tests, would just tuning to native be enough? Cause then the compiler wouldn't emit the failing instruction, right? I haven't tried that yet.

<kestrelwx>I guess it could also happen due to containers.

<ieure>Could it?

<kestrelwx>ieure: I used to see a `SIGILL` with regards to GL or Vulkan drivers in a `guix shell -C` environment, althought I haven't bothered debugging. So it's possible in some way.

<civodul>yelninei: small fixes you might be interested in 👉 https://codeberg.org/guix/guix/pulls/7612

<Guest27>ieure Im back because the kernel module does not get loaded

<Guest27>Config file is the same as yesterday

<Guest27>Paste.debian.net/plainh/9e85e65f

<Guest27>Except this line now reads (use-modules (gnu) (gnu packages filesystems))

<Guest27>But when I try mounting my bcache it says error, and suggests the kernel mod is not loaded

<Guest27>Ok here we go current config: https://paste.debian.net/plainh/0df1e7a0

<Guest27>Error from guix: module bcachefs not foind in drectory /run/booted-system/kernel/lib/modules/6.18.19-gnu

<Guest27>I just reconfigured just in case, no improvement

<Guest27>If anyone can help I'd be so grateful

<los7432>On guixsd, I have a root user and a main user. I don't use the root user much. So my main user's guix is on the latest commits but my root user's guix is a few months old. Should I update my root user's guix also? What's the rationale for keeping both in sync or not keeping them in sync?

<Guest27>Even if I disconnect I wil check the logs if any of you answers

<identity>los7432: first of all, it is not called GuixSD anymore, just Guix System. second, there is no point in upgrading root's Guix on System

<csantosb>Hi Guix ! Could someone do a `./pre-inst-env guix build --no-grafts --no-substitutes icecat` in current master ?

<csantosb>I'm getting a hash mismatch with boost@1.83

<los7432>identity: thanks. that's what I wanted to hear (one less thing to do). i was curious if there might be a valid reason beyond my knowledge. you don't know what you don't know

<identity>los7432: on a foreign distribution you need to upgrade root's Guix to upgrade the daemon

<los7432>one reason i can think of is if someone has a lot of user accounts on the system and to keep everyone's individual guix updated without duplication, you setup a shared cache service and expose the root user's checkouts

<identity>not necessary to involve root in that

<Guest27>Concerning loading dkms i think they dont load if not included in kernel module loader service type

<los7432>identity: all the users package definitions will be pulled from the root right? that would necessitate that root guix remain up to date?

<identity>los7432: you can use any user's cache for that, though, not necessarily the root user's

<los7432>ah right got it

<chris0ax>hey people, ive been trying to make a one-shot shepherd service to start swayidle on boot. But when running it that way, its not able to get the WAYLAND_DISPLAY environment variable and errors out. Any idea what else I could do?

<identity>and i am pretty sure you still have to pull as every user, you will only skip the git-pull part of it

<identity>chris0ax: make your compositor start it?

<identity>both niri and sway have a setting to execute a file on launch, at least

<chris0ax>identity: ohhh, that would solve it niri does have a 'spawn-at-startup' option..ill try that, thanks!

<Guest27>Hey basic question

<Guest27>When I wanna add a service

<Guest27>Do I use the syntact services (cons* (service

<Guest27>Or the syntax serbixes (append

<Guest27>In the manual they use one and the other

<ieure>Guest27, It depends on what you're adding and your personal taste. They work the same.

<ieure>I mean, they accomplish the same result.

<ieure>I generally lean to cons* because it doesn't require a nested list.

<Guest27>Oh I see thanks ieure let me try

<Guest27>What would be the name of the srrvice which corresponds to the module bcachefs-linix-module?

<ieure>I don't believe there is such a service, what makes you think there is?

<Guest27>ieure im trying to use kernel loader service type to load bcachefs-linux-module

<Guest27>The name of the package is bcachefs-linux-module

<identity>yeah?

<Guest27>But in the examples Ive sesn thats not what they write together with loader servixe type. Eg

<Guest27>Eg from the guix manual

<Guest27>kernel-loadable-modules (list ddcci-driver-linux)))

<Guest27>But then they write services (cons* (service kernel-module-loader-service-type

<Guest27> '("ddcci" "ddcci_backlight")) ieure

<Guest27>So in my case I have bcachefs-linux-module and I wonder what variation of that I shiuld put in the " "

<identity>guessing: "bcachefs"

<Guest27>Lets see if this shit works

<Guest27>No

<Guest27>I must have made a syntax error

<Guest27>Can someone take a look at the syntax at lline kernel-module-loader service type? https://paste.debian.net/plainh/7424ab13

<ieure>Guest27, Please see the "Kernel Module Loader Service" portion og the Guix manual, which has examples.

<ieure>Guest27, You can also look at this code from my personal channel, which configures loading of the ZFS module: https://codeberg.org/ieure/atomized-guix/src/branch/main/atomized/system/profiles.scm#L471

<Guest27>Eh I know ieure but the networkmanaher examples use append and the kernel module loader exmaples use con* thats why I am struggling

<ieure>Guest27, (cons* 'foo 'bar '(baz quux)) -> '(foo bar baz quux)

<ieure>Guest27, (append '(foo bar) '(baz quux)) -> '(foo bar baz quux)

<ieure>They're just two ways of adding elements to a list, cons* works best if you want to define the things to add in-line with the cons*, append works best if you already have a list of them.

<identity>also, (cons* 'foo 'bar '(baz quux)) expands to (cons 'foo (cons 'bar '(baz quux)))

<squid64>Hello

<squid64>I did a guix system reconfigure and when I try to log into GNOME with gdm I get "authentication error" but if I try hyprland and remove GNOME I can log in just fine. Any idea?

<squid64>I thought it was a gdm problem at first but I guess not if I can log in just fine on something else.

<Guest27>ieure im tempted to say there is sth wrong with the ' but Im not sure.

<ieure>Guest27, With what quote?

<Guest27>I mean the line (service-kernel-loader etcetc) should have the same syntax regardless of whether i use append or const

<Guest27>Yeah that lonely quote is fishy

<Guest27>But that inner parenthesis is copied from the examples and has the lonely quote so shiudld be xorrect I expect

<Guest27>I mean the probl should be on that one line of code

<Guest27>Irrespective of whether i use cons or append

<Guest27>But that line of code is copied from the manual

<Guest27>Hence shiuld be correct

<Guest27>So im at loss

<bjc>can you paste your config at paste.debian.net?

<identity>Guest27: just checking, do you know Scheme? have you read (info "(guix-cookbook) Scheme tutorials") ?

<Guest27>I dont know scheme! But until i setup my laptop i cant learn it. Im messaging from my phone now

<ieure>bjc, Config is already pasted / in scrollback.

<bjc>oops

<bjc>i'm getting: /tmp/a/cfg.scm:63:32: error: (uuid "f65" (quote dce)): invalid UUID

<bjc>which, yes, it is an invalid uuid

<bjc>that's from your swap line

<Guest27>bjc are you talking to me?

<bjc>yes

<bjc>you can use `sudo blkid` to get the uuids you need

<bjc>you can also use labels if you'd prefer that

<Guest27>bjc I anonymised the IDs in the paste

<Guest27>The problem is in the line that has the service kernel loader

<Guest27>The syntax there must be wrong bjc

<identity>«The syntax there must be wrong» why?

<Guest27>bjci get unbound variable error

<bjc>looks fine. if the syntax were wrong i woul'nt have this error

<bjc>ah

<identity>you should have said that waaaaaay earlier

<Guest27>identity i get unbound variable sth is wrong there

<bjc>add: (gnu services) to your `use-modules` line

<Guest27>bjc that solves the unbound var?

<bjc>should

<identity>if you get an error you better tell us what the error is, we can not materialize the error message in our minds

<Guest27>So its not a syntax error

<Guest27>Many apologies

<bjc>np

<bjc>how would you know?

<Guest27>I understand I wasted some of your time which is worth a lot and a lot of mine which is worth medium

<Guest27>Sorry

<bjc>`guix system search` is useful for finding where things are, if they're not in the manual

<Guest27>Ok let me try again now

<bjc>finding where system services are, that is. `home search` for home services, and plain old `guix search` for packages

<identity>i would not say our time is definitely worth more, but there is more of our time than of your time because there is more of us than of yous

<bjc>we're choosing to spend time here to help when we can. it's a silly thing to argue what's worth more

<identity>indeed

<Guest27>identity doesnt wlrk

<Guest27>Work

<Guest27>Same error:

<Guest27>Guix/ui.scm:920:18: error: kernel-modile-loader-service-type: unbound vRiable bjc

<identity>what does the error say

<identity>you need to use the module that defines kernel-module-loader-service-type

<Guest27>I added (gnu services)

<Guest27>So its not that one

<identity>yeah, not that one

<Guest27>Hum

<identity>what is the section of the manual that describes kernel-module-loader-service-type called?

<Guest27>Maybe (use-package-modules linux)

<Guest27>(use-service-modules linux)

<identity>right

<bjc>yeah, that one too

<Guest27>Shall I add it and see how it goes?

<bjc>it's already building for me, so i have high hopes

<identity>go ahead

<Guest27>I dont understand if the too is referred to the working or to the not working

<bjc>you'll need both imports, the (gnu services) one you've already added, as well as adding ‘linux’ to the ‘use-service-modules’ line

<Guest27>And the use package modules? Also relevant in your opinion?

<bjc>won't hurt, and it's in the documentation

<bjc>so, you know, just do what the docs say =)

<ieure>I believe the bcachefs package is defined in (gnu packages linux).

<bjc>it's a string, though, not the module package

<identity>bjc: pointless, (gnu) re-exports (gnu services)

<identity>and seemingly half of everything

<bjc>documentation patches welcome ;)

<squid64>Is there another good desktop like GNOME that would work well with Guix? Like one that has audio and everything else and lockscreen, wallpaper, etc out of the box. Because this is just not working.

<squid64>A wayland desktop

<bjc>i use both kde and gnome on guix

<identity>squid64: try KDE or (not sure about Wayland) MATE

<bjc>both wayland

<identity>or whatever desktop the installer offers, really

<identity>as in, any of

<squid64>I see because yeah GNOME doesn't work after updating anymore.

<bjc>does it do more than gnome now?

<squid64>I'll try KDE

<identity>bjc: GNOME, Xdce, MATE, KDE Plasma are the DEs (it also offers a few X11 WMs and a Wayland compositor)

<identity>Xfce

<squid64>does xfce has wayland support now?

<bjc>oh yeah, and it's always offered emacs wm

<squid64>I would love exwm tbh

<ieure>squid64, I don't know how well it works, but there are both Wayland and X11 options for xfce in gdm.

<squid64>But no wayland support on exwm

<ieure>No.

<ieure>I still use X11 and EXWM. Wayland has nothing I care about.

<squid64>I haven't found any exwm wayland alternatives either. I hope there is one eventually.

<ieure>squid64, EWM and reka are two EXWM-like options for Wayland, but both are materially worse (ex. no simulation keys).

<identity>no simulation keys /for now/, for reka at least

<squid64>I see. First time I hear of those, good to know there is something and it's not yet another dead project right?

<identity>not yet

<ieure>lol, I mean, EXWM itself is barely alive.

<squid64>Yeah but exwm is pretty much complete for X11 already so it's good enough for X11.

<identity>whether X11 is good enough is a different question

<ieure>Yeah. I've been daily driving it for like... eight years now? I think around tha tlong.

<squid64>Yeah X11 has been abandoned and more and more people are moving to wayland so I think in the future there will be issues with running the latest software on X11.

<squid64>I heard of xlibre too but I haven't tried it and I heard it's worse than xorg

<identity>apart from, uh, other problems

<squid64>GNOME will drop or has dropped xorg support and KDE plans to in their next release

<squid64>ok kde has been installed, lets see if it works

<identity>will or did depends on what time is it. Guix packaging is an alternative timeline

<squid64>Yeah guix doesn't have the latest of those. So GNOME and KDE users who like xorg will get a bit more life out of it

<identity>will GDM still be able to start X11 sessions in 49+?

<squid64>No sure, but I wouldn't be surprised if it no longer can.

<squid64>I know gdm will be wayland only though but whether it will be able to run X11 sessions, not sure.

<squid64>Also KDE works yay

<identity>according to a random website, they will keep the functionality in GDM and enabled

<ieure>There are other display managers.

<identity>news to me

<ieure>Is it?

<identity>GDM is the default in Guix, though

<squid64>Yeah, gdm is part of desktop isn't it

<ieure>Yes, but you can change that, like pretty much everything else.

<ieure>sddm is another option that's ready to go in Guix.

<squid64>sddm is pretty good as well yeah

<squid64>Anyway, guix is pretty nice other than GNOME no longer wanting to work for me lol

<squid64>I like that changin desktops is easy and that you can easily just delete the old one and replace it, same as on NixOS

<hwpplayer1>Do you have a programming language developer policy ? I mean we accept C but not C++ something like that

<identity>WDYM?

<hwpplayer1>Do you accept developers in guix distribution who write in any technology with a free software license or only accepted technologies /

<hwpplayer1>?

<hwpplayer1>Do you accept C++ for instance

<ieure>hwpplayer1, It has to have a Free license and compile from source.

<hwpplayer1>So any Free Tech is acceptable right ?

<hwpplayer1>Not only Lisp based

<identity>why would this be a question?

<ieure>hwpplayer1, As I said, Free and can compiles from source. If we only accepted software written in Lisp, how would we run Linux?

<hwpplayer1>I am just curious and making a R&D

<ieure>We do not have any policy on accepted/rejected languages, and I'd find that super weird in any distro.

<ieure>hwpplayer1, Your curiosity is trivially satisfied by examining the packages in Guix.

<hwpplayer1>Okay

<ieure>Or even the Guix source code itself, which has some C source.

<hwpplayer1>I see that today

<identity>the daemon is still in C++

<hwpplayer1>.cc

<hwpplayer1>Okay

<hwpplayer1>Thanks

<ieure>hwpplayer1, What do you want to accomplish that led to this question?

<hwpplayer1>I want to know the tech and mindset behind GNU/Guix/Linux and Hurd

<hwpplayer1>Just loudly thinking

<squid64>Lisp is a pretty nice language so far for me at least. Coming from Haskell.

<hwpplayer1>I was not aware of this

<squid64>I only know emacs lisp though, but scheme isn't too different.

<hwpplayer1>Thanks for the information

<ieure>It's pretty different, especially when it comes to stuff like macros and nil punning.

<ieure>I miss nil punning so much.

<squid64>there's no nil?

<ieure>I did not say that *at all*.

<squid64>what's nil punning?

<ieure>It's the ability to treat nil as an empty collection of whatever type a function requires. So things like (car nil) evaluates to nil.

<squid64>Oh I see

<ieure>Scheme doesn't let you do that, so if you wan to chain together a bunch of functions that handle a possibly-nil value, you have to litter it with guards.

<hwpplayer1>Great to see a good and useful conversation as always again

<ieure>Also many functions return #f instead of #nil, which also require guards.

<ieure>I find Scheme to be unpleasantly fiddly vs. other Lisps I know (Emacs Lisp, Clojure, SBCL/Common Lisp).

<squid64>Are guards in lisp the same as guards in Haskell?

<squid64>As in they act like if else statement

<ieure>squid64, "Guard" isn't a specific, named language feature. You need some kind of conditional, there are options, including `ife.'

<ieure>Sorry, `if'.

<ieure>and=> is the tidiest way to get a similar effect... But I like that you don't need that kind of thing (for this) in other Lisps.

<squid64>I see. I am pretty new to the language but I think it's cool. especially how functions are used.

<squid64>I also always wonder why people like lisp so much, like they'll write everything in the language.

<squid64>It got me curious to learn more about it

<Guest27>bjc it still doesnt work. Mount bcache says bcache module not loaded? How can i check what modules are loaded on my system and what do I need to add to my config at this point? Config current: https://paste.debian.net/plainh/633651e1

<Guest27>I am disconnecting to go to the supermarket but will connect when back and will check the logs

<Guest27>ieure any help appreciated

<bjc>`lsmod` will tell you what modules are loaded

<Guest27>Thanks bjc so bcache is not loaded

<Guest27>Bcachefs

<mra>howdy guix o/

<mra>currently in the middle of building binutils-mesboot. looking like it's going to take a hot minute

<ekaitz>hehe

<ekaitz>minute

<ekaitz>mra: did it finish?

<mra>ekaitz: seems to have! building gcc-4.6.4 now

<ekaitz>oh!

<bjc>oh man, you're doing the whole thing from bootstrap?

<mra>yep

<ekaitz>in x86 is faster than in RISC-V that's for sure lol

<mra>gave up on figuring out whatever the hell weird nixos bug was causing my guix installation to fail, so i'm stuck without substitutes for the time being

<bjc>i think the last time i had to do that it took about a day or two, but that was on relatively modern hardware

<kestrelwx>mra: What do you get with `guix describe`?

<kestrelwx>My friends on Nix have had issues with substitutes and seen `#f` in describe.

<mra>kestrelwx: #f for everything

<ieure>kestrelwx, I believe that's expected if you install Guix on a foreign distro and have never run `guix pull'.

<mra>bjc: yeah, it's been running overnight and it's still not done

<mra>making progress though

<bjc>gcc 4 is still pretty early, unfortunately

<bjc>there are so many gcc compiles

<kestrelwx>ieure: makes sense.

IRC channel logs

2026-04-01.log