IRC channel logs


back to list of logs

<below_turtles_>how to extract a nar bundle or split one up to pass to guix archive -x?
<graywolf>Hi Guix, if I want to send a patch for the python package itself, how should that be done? git send-email --subject-prefix='PATCH python'? python-team?
<graywolf>I assume it should not go on the master branch...
<graywolf>Hm, let's go with python-team.
<podiki>graywolf: yeah python team
<podiki>if it is for new python version, that could go to master, just versioned (like python-3.11 or whatever); switching the default to a new version would be done separately definitely
<podiki>and will need testing/fixing for packages i bet
<apteryx>graywolf: even if you forget the python suffix, members of that team shoud be notified via 'git send-email'
<graywolf>podiki: not a new version, I finally found a time to make python build reproducible
<podiki>ah nice
<podiki>yeah, that will have to go on a branch since it will rebuild the world
<apteryx>core-updates sounds like a good place -- if civodul agrees to let it in
<apteryx>seems like it should be doable if it's a small patch fixing just that problem
<graywolf> the patch is pretty small
<peanuts>"[PATCH python-team] gnu: python: Make the build reproducible."
<graywolf>well I am off to bed, it is past 2am over here; have a good whatever is in your TZ right now
<roran>who made steam not build ;-;
<bdunahu>Hi, I am trying to fix my grub configuration, though whenever I chroot into my system, I seem to lose internet (even with ethernet). Does anyone have experience with this?
<bdunahu>I am using section 12.1 of the manual and have tried what was listed there
<below_turtles_>best way to download and find all the sources in the store required to build a system no substitutes?
<apteryx>which package provides 'dmd' ?
<apteryx>I thought it would have been ldc
<apteryx>but it's not
<apteryx>it has 'ldmd2' though
<podiki>roran: actually it's from guix in this case ;) check2 failing a test on i686
<bdunahu>I found the solution to my previous question, so nevermind that
<roran>podiki: does the manual cover how i can temporarily mask that? i guess i can just remove the package
<ryblade>this could just be the caffeine talking but damn i'm enjoying learning how to use guix and scheme, it's fun :)
<roran>same, ryblade
<jaft>'Dunno if anyone would know but I'm trying to use the MPD service; 'was trying to run it as myself so I used ~(service mpd-service-type (mpd-configuration (user jaft) (endpoints '("localhost"))))~. It runs but, if MPD gets used first (on boot), no other application is able to use the audio; but, if any other application runs first, MPD is unable to use the audio.
<sneek>jaft, you have 1 message!
<sneek>jaft, lilyp says: desktop-read is a function that reads project-like buffer configuration from a file – pretty basic, though perhaps a bit arcane emacs stuff
<villageidiot>Hiya folks, for the Wayland/Sway/swaylock users: swaylock won't accept my password and I'm wondering why. I checked the debug output and can see that PAM authentication fails due to "invalid credentials". This seems similar to this issue, do I need to add some PAM rules manually?
<peanuts>"swaylock fails to unlock ? Issue #3631 ? swaywm/sway ? GitHub"
<villageidiot>I ask because I'm hesitant to write any system-level config files by hand in Guix. It wouldn't feel right.
<villageidiot>Also, on the topic of PAM, I asked here the other day but I didn't end up figuring out how to add PAM rules in my system config. I had a look at (gnu services pam-mount) for hints to no avail.
<podiki>roran: masked? you can roll-back, time-travel
<adanska>Hi Guix!
<ryblade>villageidiot: i tried looking for some documentation regarding PAM configuration, unfortunately this is all i found:
<peanuts>"bug#37583: Documentation needed: pam-services in operating-system struct"
<ryblade>it's a shame because i've been quite impressed with guix's documentation for the most part
<ryblade>i knew i'd run into something today... here's my problem: 'guix system reconfigure' failed to build the 'acl' package. since i don't need that package anymore, i removed it from my packages list in /etc/config.scm, but 'guix system reconfigure' still complains that 'acl' won't build.
<villageidiot>ryblade: thanks for the info, at least now I know there is such a thing as `pam-services`. I'll try to look further.
<ryblade>well, i did it again. set a goal to make a quick little backup script, several hours later and i've written a small frontend in bash with selection prompts and arguments. i should sleep.
<roran>ryblade: any code that I can see?
<ryblade>roran: nah, it's completely untested... i may share it if it works, but i'm not trying it out until i've rested. remind me later.
<roran>sounds good; good luck.
<civodul>Hello Guix!
<futurile>Morning all
<futurile>Q: I've built a package - how can I see whether a patch was applied to the build?
<futurile>I found package-direct-sources, but I can't figure out how to get from the build (I have the drv and the package) back to the package. Or any other way of seeing the patches that were applied
<civodul>futurile: you can look at ‘guix build -S PKG --log-file’, for instance
<civodul>that’s the log file of the source
<futurile>civodul: I can see the patches being applied in the there? is there a patch phase?
<civodul>futurile: it’s not a phase strictly speaking, but you should see lines like “applying patch …” in the log
<futurile>OK - going to play with something in guix where patches must work - as I can't grep/see that in mine
<rekado>numba failed on master
<rekado>oh, sorry: it’s on powerpc64le-linux, not x86_64
<bost>I have two machines running Guix and I'd like to be able to establish remote desktop access between them. Is there any guide how to do it? I guess I need to `guix install rdesktop xrdp` and then ???
<futurile>bost: I don't think there's any specific guide for Guix. Any standard guide should be fine I guess.
<allana>Hi #guix! I am likely going to submit an issue based on my experiences over the past few days with inferiors and channels. Can anyone let me know if I have made false assumptions on how inferiors and channels work? I have constructed what I think is a minimal example of the problem in this paste:
<peanuts>"debian Pastezone"
<allana>Actually, when copy-and-pasting into the paste I somehow thew in a typo. This is the appropriate paste:
<peanuts>"debian Pastezone"
<bost>allana, in my the have a look at the function inferior-pkgs and the way I use it.
<peanuts>"dotfiles/guix/home/common/cfg/packages/all.scm at master ? Bost/dotfiles ? GitHub"
<allana>bost, thank you! will take a look now.
<jpoiret>civodul: I'm slowly making my way up to gnome on c-u, I need to send a couple of patches. What's the status of c-u for now? I see you've pushed a couple more low-level changes
<allana>bost: Do you use "inferior-pkgs" against any channels other than "guix"?
<bost>allana, no
<bost>allana, FYI your question reminded me that I've been using an inferior version of libreoffice. I just changed that ;-)
<civodul>jpoiret: hey! essentially, i’d be happy to leave ‘core-updates’ in its current state:
<peanuts>"An update on ???core-updates???"
<civodul>howdy bost :-)
<allana>bost: ha, glad that this helped you find that :-)
<allana>So, I'm curious to know if anyone has an example of an inferior package from a channel that's not from the "guix" channel
<emru>hey, download link to guix hurd is not working
<oriansj>it appears the link hasn't been working for a while
<oriansj>but it is probably best to do guix system image --image-type=hurd-raw gnu/system/examples/bare-hurd.tmpl to get a system image to play with
<apoorv569>I just wrote this home-xdg-user-dirs-service-type,
<apoorv569>it does work, but the config file that is generated is wrong..
<apoorv569>I tested this inside guix container and the first time the container starts these are the values,
<apoorv569>but if I like manually try to run `xdg-user-dirs-update` command (not using the shepherd service) it sets all dirs wrong..
<apoorv569>is this normal? I though that command is suppose to read the `user-dirs.dirs` file I generated.
<mekeor>apoorv569: wild stupid guess: try "$HOME" rather than "$HOME/"
<apoorv569>in the generate-xdg-user-dirs-config-file?
<apoorv569>let me try..
<jpoiret>civodul: I'll reply to that, I can lead the stabilization effort
<apoorv569>mekeor: OK, so now the initial config generated is wrong,
<apoorv569>but running xdg-user-dirs-update command works
<mekeor>looks like a bug in home-xdg-user-dirs-configuration-{desktop,...}
<mekeor>i guess the trailing slash needs to be removed
<apoorv569>According to arch wiki,
<apoorv569>this command generates thse files..
<peanuts>"XDG user directories - ArchWiki"
<apoorv569>which trailing slash?
<mekeor>apoorv569: i don't use guix-home but why do you redefine all those configuration records and services rather than just passing a custom config?
<snape>allana: here's an example with nonguix
<peanuts>"debian Pastezone"
<janneke>ACTION wrestles with core-updates
<villageidiot>Hey! Probably 10-12 hours ago I asked about writing PAM rules with Guix (for the 2nd time) and I finally figured it out after much floundering. Specifically, I needed to add a PAM rule so that Swaylock could actually authenticate when entering the correct password and unlock my screen. Long story short I extended `pam-root-service-type` with the
<villageidiot>default `(unix-pam-service)` and a transformer function that simply did nothing to the other PAM services. I'm posting the service-type here to get feedback:
<peanuts>"View paste TWUQ"
<villageidiot>Surprisingly, this worked! Glad I finally wrote my first service. But I'm wondering, can a package provide such a service? Would it not make sense to have something like this included with the swaylock package by default?
<rekado>packages cannot provide services
<villageidiot>Sorry, bad wording. The service I wrote is really just a way to install a PAM rules file. It seems more straightforward to just install that file (/etc/pam.d/swaylock) whenever you install swaylock.
<dariqq>hi, i am trying to subscribe to guix-devel and with both the web interface and mailing to guix-devel-join I didnt get any reply from the list.
<villageidiot>The only downside I can think of is that different people may want other means to authenticate with swaylock. I guess that's what package transformations are for, or maybe some other mechanism (I'm a noob at this Guix stuff).
<Guest8>I put out a simple patch that has been waiting for a while now... any committer here available to take a look? Thanks in advance.
<peanuts>"[PATCH] gnu: python-graph-tool: Update to 2.58."
<janneke>ACTION is afraid the new boost build hangs for the hurd
<snape>Guest8, please can you do a patch for 2.59?
<bost>civodul, hi
<apteryx>why is our gdc package lacking a gphobos lib?
<apteryx>that's the standard D library
<apteryx>it's missing --enable-phobos at GCC build time
<apteryx>ACTION rebuids GDC
<allana>snape: thanks, just got around to looking at your example. I get a similar error when trying to adapt it to a manifest.scm -- for me this does not work.
<snape>allana, can you show your code please?
<allana>yes! I have a paste:
<peanuts>"debian Pastezone"
<snape>but allana, guile is not a channel!
<snape>or is it?
<allana>Yes, from a blog post linked in the paste
<snape>ok mb
<jakef>is git-checkout just a lazy git-fetch?
<sneek>Welcome back jakef, you have 1 message!
<sneek>jakef, apteryx says: that's the example given in: info '(gnupg) Agent Examples'; it seems to be so that you set SSH_AUTH_SOCK only if it wasn't already for the current session (multiple login)
<bost>snape, yes it is a channel
<snape>and, allana, what is the error you get?
<bost>`sudo guix system --verbosity=3 reconfigure ...` freezes after "guix system: bootloader successfully installed on '(/boot/efi)'". Does anybody have an idea what to do?
<bost>Also `sudo dmesg --ctime` shows:
<bost>[Fri Jan 12 15:01:13 2024] ACPI BIOS Error (bug): Could not resolve symbol [^^^^NPCF.ACBT], AE_NOT_FOUND (20230628/psargs-330)
<bost>[Fri Jan 12 15:01:13 2024] ACPI Error: Aborting method \_SB.PC00.LPCB.EC0._Q83 due to previous error (AE_NOT_FOUND) (20230628/psparse-529)
<snape>oh it's written on the paste
<allana>snape: sorry, I am mostly afk. Thanks very much for having a look!
<snape>i'm trying to run it, for now it works, lets see
<apteryx>OK, it appears GDC *has* std, but it's statically linked so, there's no
<apteryx>it's part of, I tInk
<Guest8>snape: I'm planning to do that once this one goes through. I'd like for both to be available in the history. If that's not too much trouble.
<snape>oh ok np
<snape>allana, your sample works for me
<snape>maybe you have an old guix?
<snape>maybe guix pull, upgrade your guix daemon, reboot and try again?
<allana>Did yours work? I guix pull multiple times per week.
<allana>Ah, sorry I see that it works
<janneke>something's terribly broken wrt offloading on my side, downloading built substitutes just hangs
<janneke>guix copy works
<allana>Well now I am definitely lost. But at least I know there's something wrong with my setup and not something fundamental to guix. Thanks snape.
<snape>my pleasure
<janneke>hmm, guix offload test runs fine
<snape>Guest8: your patch doesn't apply
<snape>seems there is some space offset
<allana>I think my env var "GUIX_PACKAGE_PATH" was at least part of the problem. Now the example builds :-) -- Thanks again snape
<snape>nice allana, you're welcome!
<freakingpenguin>Hey all, want to check my understanding. In guix build, is the difference between --target and --system that target uses cross compilation while system emulates the build tools in QEMU?
<allana>ACTION is practically doing cartwheels now, so happy to have finally gotten inferiors to work with external channels!
<Evil_Bob>good afternoon, i have a question, can someone update the sfeed package from 1.6 to 2.0? (I'm the author) Thank you in advance
<peanuts>"suckless.scm\packages\gnu - guix.git - GNU Guix and GNU Guix System"
<Guest8>snape: I don't get it. I sent the e-mail generated by git "as is".
<snape>but your code while you generated it, was it "as is"?
<Guest8>are you not confusing the patch with this other one ?
<peanuts>"[PATCH] gnu: python-graph-tool: Update to 2.58."
<Guest8>I sent that one first, and I believe there was indeed some formatting issue there, introduced by the email client (thunderbird).
<snape>Guest8: I'm talking about this one: there's a space before (define-public
<peanuts>"[PATCH] gnu: python-graph-tool: Update to 2.58."
<Guest8>I don't get it. git would have produced a diff that shows that, right? Is it the e-mail client again? The patch was attached, I didn't touch it at all.
<snape>maybe it's the email client if you didn't use "git send-email"
<Guest8>Ok, I didn't send it as an attachment, but the e-mail I see in my sent folder doesn't have any spurious spaces.
<Guest8>I'm remembering now, I actually tried sending other stuff as attachments and that didn't go well.
<Guest8>About git send-email, since I already have an e-mail client setup, I'd rather not buy into even more stuff that I only use once every couple of months to send a few lines to Guix.
<snape>Evil_Bob: done
<Guest8>Anyway, I'm investigating to see if I can fix this. It would be simpler if it could just attach the patches.
<Guest8>s/it could/one could
<snape>best is to use git send-email, then to attach the patches
<snape>but the worst of all is to mix something up^^
<snape>to me it's the same if you attach the patches
<snape>same workflow
<snape>as long as it applies
<Guest8>Btw, how can I close the patches that have issues ? There's no indication.
<Guest8>Attaching the patches wouldn't require me setting up something else. But it didn't work last time I tried. Maybe because I put attached to patches to the same e-mail?
<cbaines>attaching patches can sometimes work, but you can't attach more than one patch to an email
<cbaines>there's also some threading and subject stuff that git send-email takes care of
<snape>Guest8: you can mail <bug id>
<Guest8>cbaines: I had a typo, I indeed meant "attached two patches", so as you say it's not supposed to work. Th
<peanuts>"Developers' information"
<Guest8>There could be something in the documentation on how to properly send patches without setting up git send-email.
<snape>cbaines, why do you say it's not possible to attach more than one patch?
<attila_lendvai>if i try to renice a process in top (procps) then backspace doesn't work, __but only in top__, and only when i'm in an ssh session. it's between two guix machines, and no weird config that i know of. is this only me?
<snape>Guest8, the reality is: git send-email is easy to use
<snape>way more than anything else around
<snape>for people who send patches, and for reviewers as well
<podiki>you can send an email with an attached patch; probably one per email is best
<podiki>but that's what i do majority of the time actually
<snape>sending several patches per email does work anyway
<podiki>plain text email, patch from format patches
<apteryx>haha, was already there in gdc, but it's in its "lib" output
<Guest8>sending several patches attached to one e-mail doesn't work; or at least it didn't in this case:
<peanuts>"[PATCH] gnu: Fixes for sequeler"
<podiki>might be a mumi thing, I see 2 patches on debbugs:
<peanuts>"#67636 - [PATCH] gnu: Fixes for sequeler - GNU bug report logs"
<snape>Guest8: I can seamlessly apply patches from
<peanuts>"[PATCH] gnu: Fixes for sequeler"
<snape>it does work
<snape>probably a mumi bug as podiki says
<Guest8>Ok, cool. So can I just ignore "mumi" and attach all patches to one e-mail?
<snape>it depends on the reviewer ^^  git send-email works with all reviewers
<podiki>i would still do one per email since most like mumi, but whatever helps gets patches in :)
<snape>with me it doesn't matter as long as it applies
<podiki>but manually sending is fine, honestly i do that almost all the time
<snape>because from emacs I can chose which patch I apply from any debbugs
<podiki>many will get patches via emacs debbugs interface, so that shouldn't matter
<Guest8>I appreciate the ease of using git send-email, but some people who contribute very occasionally (me) might not be wanting to setup another smtp server and have their email passwords in yet another file just to send two lines of code every two months...
<snape>there is no password to store
<snape>it's interactie
<peanuts>"debian Pastezone"
<snape>that's all there is to set up
<snape>there are some more in-depth examples in the guix info manual
<snape>for when you need to send a cover letter first and then wait for the reply and send patches to the reply
<podiki>depends on how you send email though, but really you don't have to, plain email with attachment is fine
<podiki>seriously, whatever lowers the barrier for patches if that's the case for you
<Evil_Bob>snape: thank you! :)
<Guest8>Mmm. I assume that's not the case with oauth2 smtp authentication. But, I guess I could create a random e-mail account for Guix, somewhere gratis that accepts password smtp auth.
<snape>yeah I agree with podiki, do attachments are good enough
<Guest8>I'm thinking that there could be some more straightforward explanation on the docs about: (a) for beginners, sending format-patch outputs as simple attachments is fine despite what you'll see in 'mumi' or the QA stuff which may also be affected, and explaining what to put in the title of the e-mail (does it matter?); (b) show how to set up git
<Guest8>send-email, or a link to some page that already explains that.
<Guest8>Is the doc part of the guix repo?
<Guest8>snape, podiki, cbaines: thanks for all the help and patience :)
<snape>I don't think there is a need for rules about everything.
<snape>and typing "git send-email setup" on your favourite search engine gives the answer already
<snape>the doc is part of the repo yes
<podiki> at the bottom about basic sending and then next sections for git send mail
<peanuts>"Submitting Patches (GNU Guix Reference Manual)"
<Guest8>Thanks. I was looking at that page, that's why I asked.
<mwette>Am I using substitute-keyword-arguments correctly here?
<peanuts>"debian Pastezone"
<jpoiret>mwette: make-flags is a either a g-exp of a list or a quoted list, so you'll need one more level of quotation in its replacement
<podiki>i think so at a quick look? but not sure why you have simple-format, just string-append would be fine no?
<jpoiret>you probably want to backquote (cons* .), then unquote dtb-path and flags
<jpoiret>(and abi and isa)
<Guest8>Rules/guidelines can make things a lot simpler for people who don't understand the whole setup. In the past year or two I have read all of the contributing docs more than once and, as you can see, I still mess things up and feel a lot of uncertainty, because I'm only doing this sporadically.
<snape>but if there was a rule that forbids the use of attachments, you wouldn't be able to send a patch right now, whereas thanks to the absence of rule, you can
<Guest8>(taken together with potentially misleading stuff like imperfections in mumi and debbugs)
<Guest8>sure, I think I misread you, sorry
<snape>a rule saying "people can attach patches" would discourage the use of git send-email, so it might not be a good idea either, etc
<Guest8>Was thinking about rules as some more explicit guidance about what works and what doesn't. As in attaching is much safer than opening the '.eml' and sending it. Anyway the docs already do a good job and I do get patches right most of the time ;-)
<snape>ok cool ^^  anyway if you feel it can be improved by adding something about attachments, don't hesitate to send a patch to fix the doc
<snape>i have no strong opinion, and you might be wright
<Guest8>I don't think you need to worry about descouraging git send-email if you present things right. It is obviously way more convenient once someone starts contributing more or less regularly. But to for people in my (current, hopefully transient) situation it would be nice to know what works and what doesn't without it.
<mwette>jpoiret: I get it. Thanks.
<snape>yeah but it's not binary, some things work better than other, but those things still work...
<snape>"sending a mail to guix-devel saying, to fix this bug you can replace this with this" can be a valuable contribution, despite not in the form of a patch
<snape>anyway I've got to go
<mwette>podiki: (simple-format #f fmt .args) generates a string. It's simpler than string-append in this case, IMO.
<snape>enjoy your night
<Guest8>well, now I'm gonna fix the mess I left with all those patches ^^
<Guest8>you too
<snape>sure, cc me and I'll review
<jpoiret>mwette: why not use (format ...) though?
<jpoiret>hmmmm, is gcc-mesboot really supposed to take hours upon hours
<mwette>I'm used to using simple-format (i.e., not importing (ice-9 format)). Been coding guile for a pretty long time now. Not so much guix, though.
<jpoiret>format is available by default without importing (ice-9 format), it's aliased to simple-format until you import it
<mwette>Ah, right. I need to break that habit. Always kinda bugged me that simple-format uses ~A vs format ~a.
<apteryx>~a works as well with simple-format, no?
<apteryx>so there's no need for ~A, as far as I know
<apteryx>or ~S
<apteryx>is someone used to build stuff with GDC / D lang?
<apteryx>I was looking at building something that wants "dmd ... -defaultlib"; I've replaced dmd with gdmd (which is a compatibility wrapper that uses gdc under the hood instead of dmd), but the -defaultlib causes ld to fail
<apteryx>even after adjusting for the phonos default library name:
<apteryx>or stripping .so suffix, lib prefix, or both
<apteryx>(gdc-11:lib is listed in inputs, so is on LIBRARY_PATH0
<podiki>apteryx: I don't have time to follow up on this, but reported on another channel that check2 could be built with clang instead
<podiki>apteryx: pasting from other channel (linked at bottom):
<peanuts>"debian Pastezone"
<ryblade>mornin' guix, hope everyone had a good sleep
<ryblade>i've got this problem: there's a package installed through my config.scm that won't build anymore for me. when i comment it out, the package *still* attempts (and fails) to build when doing a system reconfigure.
<podiki>it could be required by something else in your config
<ryblade>perhaps, i guess i might not have noticed that. the package is 'acl' (so it's probably important), but it had it in my config.scm so i could use setfacl/getfacl.
<ryblade>hmm... i guess my next question is how do i troubleshoot or report this? been itching for a reboot after changing some kernel arguments and adding some btrfs subvolumes.
<podiki>acl? recent builds on substitute server seem good, since it has many dependents i don't think it was changed recently unless you have your own version
<podiki>can you paste the build failure log somewhere?
<ryblade>never mind, i went hunting for clues in the failure log and turns out it was entirely my own dumb fault, sorry about that. all better now!
<podiki>success for easy fix, not saying you are dumb!
<superkamiguru>I am trying to package a piece of software that uses the gnu build system, but keep running into the error "sdl/mixer_sound.c:55:10 fatal error: SDL_mixer.h: no such file or directory". I have tried including sdl/sdl2 and sdl-mixer/sdl2-mixer as inputs with no success. Oddly enough using a guix manifest with the same inputs allows me to build the
<superkamiguru>same repo under guix shell. Anyone have an idea why this might be?
<ryblade>hahaha, no worries, at least i was smart enough to realize i was being dumb, right?
<podiki>that's the spirit!
<ryblade>saved myself from the embarassment of blaming the computer, that deserves an espresso
<jpoiret>superkamiguru: can you paste the build log, esp. the configure phase?
<podiki>bye for now guix, server moving/maintenance on my end (moving to guix for server too, almost done)
<superkamiguru>jpoiret: I actually was able to get it to build by adding the following to the modified phases (took an idea from another package that uses SDL2, though I am not quite sure why it works)
<peanuts>"debian Pastezone"
<jpoiret>you shouldn't need to do that, usually that should get picked up by eg. pkg-config
<superkamiguru>hmmmm ok I will try building without it again and send the log
<superkamiguru>So the log is too long for debian paste, what would be the best way to share it?
<roran>what do you guys use for the 'nslookup' equivalent
<roran>i want to ask what "bugga" is
<ieure>Oh, if you want to query a specific nameserver, dig is the tool you want.
<ieure>host tells you what your system resolver thinks about a hostname.
<ieure>I haven't used nslookup in ages, not sure it's exactly deprecated, but it seems to be The Old Way of doing this.
<roran>what's the way i'd find the package dig is contained in?
<roran>searching with `guix search dns` leads to BIND, but is that the preferred method, where i search for something and scroll through the list, or is there some method of searching for "provisions"
<roran>interesting for multiple package outputs
<apoorv569>mekeor: you mean using `(local-file`?
<mekeor>apoorv569: i have never tried guix home but i guess something like this should work: (service home-xdg-user-directories-service-type (home-xdg-user-directories-configuration (desktop "$HOME") ...))
<mekeor>good luck!
<apoorv569>mekeor: You mean provide a value for all those fields? Why just "$HOME". My idea was to give the configuration block so people can rename the directories if they want, but I am providing defaults.
<mekeor>sorry, idk what you're actually doing
<jsbiff>So, it seems that IPv6 is technically configured and working in my Guix VM, but, that IPv4 connections are being preferred. How do I configure the networking in Guix to use IPv6 first and fallback to IPv4 when needed?
<jsbiff>I found some sites that suggested I might need to tweak how dns lookups are handled in /etc/gai.conf, but I can't actually find that file on guix. Does guix use something else that replaces gai.conf?
<poselyqualityles>apteryx: this is for the catch2 tests failing on i686:
<peanuts>"debian Pastezone"
<poselyqualityles>they pass when building with clang but not gcc
<podiki>hello again guix!
<podiki>hello again guix! (hopefully this goes through...)
<podiki>sorry for the double message, but now talking through matrix (with irc bridge) run on my guix server
<freakingpenguin>podiki: Ooooooh, fancy
<freakingpenguin>Is that using matterbridge or something else?
<podiki>would love to update our synapse package and get a service for it, but for now getting things with docker
<poselyqualityles>podiki: (fyi, just posted above about catch2 i686 build, not sure if you can see it)
<freakingpenguin>My mind was blown when I saw oci-container-service.
<podiki>maybe medium-term add in some docker-compose to system or home configuration somehow
<podiki>poselyqualityles: yeah saw it, thanks. fix sounds good but i'm not sure if there's anything else to be aware of for moving to clang
<podiki>i guess if it works that's enough, has around 45 dependents if i remember, not bad
<poselyqualityles>also opened an issue with catchorg to see whether they check i686 for their gcc builds
<podiki>but they tend to use clang in general upstream?
<podiki>because that's a good reason to switch as well, if upstream tends to clang anyway
<poselyqualityles>their CI workflow builds have gcc also. i just don't know if they target i686 at all
<apoorv569>mekeor: Looks like I have to run it with `--force`
<podiki>i686 is falling by the wayside for many distros, might have to reduce on the guix side eventually too
<apoorv569>default behavior is to reassign missing directories to $HOME
<jsbiff>one can install, using the cli:
<jsbiff>guix install bind:utils
<jsbiff>Where, of course, you are specifying the utils output of the bind package (which has multiple outputs, and the default out output doesn't install the utils like dig and nslookup).
<jsbiff>But how do you do the same thing in /etc/config.scm
<jsbiff>When I tried adding: (package->specification "bind:utils") to the list of global packages, it throws an error?
<jsbiff>That is, when I run guix system reconfigure /etc/config.scm, I get an error about that package
<rekado>jsbiff: specification->package+output
<jsbiff>rekado so, like, (specification->package+output "bind:utils") ?
<jsbiff>Or (specification->package+output "bind" "utils") ?
<jsbiff>Or something else?
<vagrantc>oh, aarch64, how I've missed you ... building /gnu/store/lj5yyyry8g2f3saxv49s5zmvbrsr5c68-linux-6.6.10.tar.xz.drv...
<vagrantc>oh no, it's also trying to build the entire chain of rust
<jsbiff>Are there no prebuilt substitutes for aarch64?
<roran>Do any of you use keepassxc across your devices?