IRC channel logs


back to list of logs

<Kabouik>Thanks podiki[m], that seems useful. I see " automatically open a useful buffer in a new frame (usually your most recently viewed buffer which isn't currently open)" and that's exactly what I was trying to do after enabling frames from the wiki above
<Kabouik>As typically I would want my terminal window with ESS ( to keep displaying just the ESS buffer, and another terminal window showing emacsclient and displaying the last created buffer, every time one is created
<arescorpio> > `guix import texlive PKG` fails with signal 11, no output
<rekado>arescorpio: this bug report has long been closed
<rekado>why did you link it here?
<podiki[m]>Kabouik: welcome! not sure how that plays for using -nw (I long ago switched to gui emacs) but hopefully something useful in there
<rekado>Kabouik: frames in GUI emacs do open new windows that your WM can manage
<rekado>is there a particular reason why you prefer to use terminal mode?
<rekado>it’s strictly less powerful than the GUI mode: no images, mangled keybindings
<rekado>if this is just about the toolbar or decorations: you can disable them.
***emyles is now known as miyulz
<imtheonewhodudes>hi guixers, in your knowledge is there any plan to support pep517 in python-build-system?
<unmatched-paren>imtheonewhodudes: There are, I think.
<unmatched-paren>It was talked about during the Guix Days event.
<imtheonewhodudes>do you happen to have some link where i can learn more about it?
<imtheonewhodudes>omg thanks a lot
<lilyp>does guix have a utility to do ".c" → ".o" suffix transformations?
<lechner>lilyp: maybe bash, for one?
<lechner>yeah, that's the next answer below
<lechner>'make' also offers such a facility
<lechner>i personally find Perl the second-easiest
<lechner>lilyp singpolyma: how about this concise version in Bash?
<lilyp>I meant Guix as in Guile + the Guile libraries used at build side
***LispyLights is now known as Aurora_v_kosmose
<phodina[m]>Hi Guix,... (full message at <>)
<podiki[m]>what's the error? maybe move the patches to outside the if and make the else part the empty list '()
<mroh>yea, perhaps you could try an empty (patches '()) in else.
<podiki[m]>(or just make the else part be (patches '()) maybe)
<podiki[m]>funny was just doing this to conditionally add a phase to do a substitute based on arch, needed the '() in the else otherwise it is trying to splice nil into the phases list (or something like that)
*podiki[m] is off, good luck
<brendyn>i really wish people would stop sending links to text messages instead of just sending the text
<roptat>hi guix!
<roptat>brendyn, it's matrix, not users
<brendyn>i know, but but i still blame them :p
<roptat>found a small issue in the dovecot service
<roptat>it reimplements mkdir-p/perms, but always sets user only for /var/run/dovecot
<roptat>that directory is hard-coded ^^'
<roptat>while investigating an issue where on one server, /var/run/dovecot is owned by root and /var/run/knot is owned by dovecot :/
<roptat>I don't get it, on another server with mostly the same config, this issue doesn't happen
<roptat>er... well, on that other server /var/run/{dovecot,knot} are both owned by knot
<roptat>somehow dovecot doesn't mind
<phodina[m]>Thanks, that also does not work. Here's the diff and error
<brendyn>phodina[m], i believe its because of your use of ,@
<brendyn>just use ,
<brendyn>you can see its creating (patches) because it ends up as (patches . ())
<brendyn>phodina[m], did it work?
<roptat>I can't remember how to run a single test, for gnu/tests/mail.scm
<roptat>make check TESTS=gnu/tests/mail.scm runs 0 tests
<phodina[m]>podiki: mroh Solved. The issue was to put the quasiquote into the patch field. Error was caused by having it around origin.
<NinjaTrappeur>Hey, it seems like Rust is bootstrapped from mrustc in Guix. I'd be curious to know how long it takes to run the whole bootstrap chain on the Guix build farm but I'm unsure where to start to get that number. How would you approach this problem?
<NinjaTrappeur>Running the bootstrap chain locally on my system is off the table :)
***roptat is now known as Guest2228
***Guest2228 is now known as roptat_
***roptat_ is now known as roptat
***roptat is now known as help
<brendyn>NinjaTrappeur, its very crude but you can look on the build servers they have the logs available.
<brendyn>and you can see the line phase `build' succeeded after 2915.8 seconds
<brendyn>you would have to go through the whole chain and add it up though
<bovid-19>hi guix!
<unmatched-paren>hello bovid-19
<Lumine>Good morning #guix
<unmatched-paren>morning! :)
<user_oreloznog>hello guix!
<bovid-19>Hello Lumine, hello user_oreloznog
<user_oreloznog>Hi bovid19, unmatched-paren, Lumine :)
<user_oreloznog>*bovid-19 :)
<NinjaTrappeur>brendyn: thanks :) I've been trying to poke the cuirass API without success to retrieve the list of the build id involved.
<brendyn>there is this data thingy
<NinjaTrappeur>Using the HTML form, I can find the build ID just fine:
<NinjaTrappeur>However, translating the same request to an API query doesn't work
<NinjaTrappeur>curl "" | jq
<NinjaTrappeur>I guess I'm doing something wrong here
<bovid-19>A while ago, my display started going to sleep after a few minutes of inactivity. I assume this was somehow automatically set up when I installed Gnome in addition to StumpwWM. I couldn't find any documentation on how to change the settings. Any ideas?
<bovid-19>That's usually fine, but it also happens when I watch videos with VLC.
<NinjaTrappeur>For the record, I found the issue: I was querying the nix name instead of the job name.
<NinjaTrappeur>curl "" | jq <= right query
<Lumine>bovid-19: ? I have no experience with StumpWM with Gnome though
<Lumine>What I do in EXWM is use xset to disable the screen blanking
<bovid-19>Lumine: Thanks, I'll have a look. I am not using Gnome anymore, I just installed it temporarily when something wasn't working with StumpWM and I didn't have the time to find a fix.
<bovid-19>Ah, sounds like that's what I should do.
<Lumine>I see
***Dynom_ is now known as Guest1775
<dgcampea>can /run or /var/run be mounted as a tmpfs ?
<Lumine>Morning sektor[m]
<nckhexen>sneek: later tell bovid-19: This is wild even as guesses go, but a ‘recent’ change in power saving could be related to <>. Before that, Guix would not use elogind defaults, despite claiming to.
<nckhexen>sneek: botsnack®
<jonsger[m]>how do you folks do screen cast (video) on Wayland on Guix?
<dthompson>jonsger[m]: does OBS work on wayland?
<dthompson>maybe a naive question...
<jonsger[m]>yeah it does work, but it seems to require for screencasting.
<jonsger[m]>to lazy ATM to package it :(
<nckhexen>dgcampea: After two failed boots: no.
<Kabouik>Anyoe using emacs EWW? Everytime I try to open an URL, it errors out saying that emacs need to be compiled with libxml2, but that's surprising if it isn't since we do have eww packages in Guix
<unmatched-paren>Kabouik: Works for me
<unmatched-paren>I'm on emacs-next-pgtk though
<Kabouik>Maybe that's because I installed emacs-minimal then indeed
<unmatched-paren>Oh, yes, it will be.
<unmatched-paren>emacs-minimal is pretty much only for emacs-build-system, afaics.
<Kabouik>It was working as a text editor so far, but I don't really know what I'm missing yet. Will try emacs-next-pgtk.
<unmatched-paren>I think just regular emacs will work, too.
<Kabouik>I prefer avoiding things like Doom for now because I'm already struggling to understand the basics and configure it, using something like that would not help
<unmatched-paren>emacs-minimal, being a minimal package, will probably be compiled without libxml or anything like that.
<Kabouik>Makes sense.
<davidl>I have a problem booting one of my systems, even after reinstalling and can't figure out whats not working. The grub boot menu comes up but then it just stops at a Guix background login screen forever. I have full disk encryption with luks 1.
<davidl>my partitioning is part1: 2 MB bios boot, part2: luks 1 and btrfs (incl boot), and part 3 is a luks 1 but a plain block device - not mapped as a filesystem.
<davidl>(actually part2 is a LVM vg)
<davidl>(I mean part3 is)
<sektor[m]>unmatched-paren: I got your note. Am a bit confused though; the left margin in a browser says no-break space but Emacspeak just says regular space.
<two[m]>$ guix shell qtbase
<two[m]>321,6 MB will be downloaded
<two[m]> qtbase-6.3.1-debug 288.2MiB
<two[m]>why does it want 288 mib of debug? i don't want debug, can i get a shell without downloading it?
***polo__ is now known as polo
<rekado>two[m]: ‘guix shell’ will get all outputs of a package
<two[m]>rekado: ```... (full message at <>)
<PotentialUser-70>is GNU Guix outdated
<unmatched-paren>PotentialUser-70: what do you mean by that?
<PotentialUser-70>I mean by the distro
<unmatched-paren>Some less-used things might be outdated, because we don't have a huge team like some more mainstream distros do
<PotentialUser-70>Oh ok
<unmatched-paren>But most packages will be up to date.
<Cairn>PotentialUser-70: Any specific packages you're worried about?
<Lumine>Taken and left
<kaelyn>rekado: last night I finally found how to fix the build of mpv with my updated vulkan and spirv packages, and sent an email to about it.
<rekado>two[m]: this is not a refutation of what I wrote.
<rekado>kaelyn: I don’t see the email yet
<kaelyn>rekado: apparently it didn't go through because the bug was archived :/
<kaelyn>basically adding "-DSPIRV_TOOLS_BUILD_STATIC=OFF" to the configure flags for spirv-tools fixes the mpv build (I found it from the Arch and Gentoo package defns for spirv-tools)
<kaelyn>(unarchived the ticket and resent the email)
<itd_>Hi. Who receives mails sent to; the reporter and subscribers? Should one Cc (all?) people that participated in the discussion? (Is there documentation that addresses such netiquette questions?) Thanks!
<vagrantc>itd_: probably just the the subscribers (unless is significantly different than
<vagrantc>i've at least observed the trend of CCing relevent people through a given bug report
<csepp>when a package doesn't build due to grafts, how is one supposed to track down where the error is coming from?
<csepp>eg.: guix build krita vs guix build --no-grafts krita. why does one succeed and not the other?
<itd_>sneek: later tell vagrantc: Thanks. :) (Not sure if it is still true, but: "Debbugs claims to provide a per-bug subscription feature [...] However, this feature doesn't work [...]" )
<sneek>Will do.
***justVain is now known as justache
<help>I'm trying to add java-brotli in java-compression.scm, but when I do that and inherit from brotli (from (gnu packages compression)), brotli is suddenly not found
<nckhexen>Cairn: To be charitable: 1.3 *is* by all standards 'out of date'. If you're not used to rolling releases, criminally so.
<PotentialUser-94>hi! after adding (service unattended-upgrade-service-type) to their config.scm and reconfiguring, did anyone else get the error that unattended-upgrade-service-type is an unbound variable? if so how did you guys resolve it?
<Kabouik>Does this work for other emacs users on Guix/Wayland?
<unmatched-paren>PotentialUser-94: Hello! Is (gnu services admin) imported?
<Kabouik>It does not for me, wondering if emacs was built with the necessary parameters
<PotentialUser-94>it most certainly is not!
<PotentialUser-94>thank you!
<PotentialUser-94>I will try it
<two[m]><rekado> "two: this is not a refutation of..." <- can i make it ignore debug?
<PotentialUser-94>unmatched-paren: it worked! may i ask how you knew to do that? i did not see anything about that in the documentation for that service
<davidl>anyone who has a working full encryption (incl. /boot) setup with btrfs? who can share their config?
<unmatched-paren>PotentialUser-94: ``guix system search''
<PotentialUser-94>ahhhh thank you
<unmatched-paren>It's ``guix search'' but for Services(tm)
<kaelyn>Kabouik: Have you tried it with emacs-next? I haven't tried transparency, but the commit linked in the post looked to be on the master branch but not the 28.1 branch of emacs.
<Kabouik>I thought 28.1 was from April and the blog post said the merge happened in January, didn't check more than that
<Kabouik>I can try emacs-next
<Nazar>    Hello! i'm trying to configure mysql service  as mysql and not mariadb
<Nazar>    and got an error like this "error: reference to invalid output
<Nazar>    'lib' of derivation
<Nazar>    '/gnu/store/swk7ydgkwqg1pmj9jmm8aivrpxr7wsvj-mysql-5.7.33.drv'"
<Nazar>    this is something that shoud be fixed right ? in default config
<Nazar>    with mariadb it works well
<Nazar>here is declaration "  (service mysql-service-type
<Nazar>          (mysql-configuration
<Nazar>               (mysql mysql)
<Nazar>               (socket "/run/mysqld/mysqld.sock")
<Nazar>               (extra-content "datadir=/warehouse/databases")
<Nazar>           (auto-upgrade? "#f")))
<Kabouik>To be honest I was thinking I'd use only emacs -nw because I want a terminal experience (with my terminal keybindings to change font size, my term colors, etc.), but then I heard it's not a good idea and I'm not doing myself a favor if I do, so now I'm trying to use emacs with its GUI but make it look like my terminal
<unmatched-paren>Nazar: please use a paste site for code/logs, eg
<Nazar>ok, thank you
<PotentialUser-94>also it seems to me there isn't a file at /etc/ssh/sshd_config as is the case in say ubuntu. is this because the idea in guix is to modify the ssh service by editing the config.scm file itself?
<Kabouik>I think yes PotentialUser-94, I went through the same questions when I configured mine
<unmatched-paren>PotentialUser-94: Yes.
<PotentialUser-94>great thanks! just wanted to make sure I wasn't missing something1
<Kabouik>Still no luck with emacs-next kaelyn, maybe I am doing something wrong
<PotentialUser-94>is chsh not the intended way to change a user's shell? i tried using usermod, but my shell reset from fish to bash on reboot
<unmatched-paren>PotentialUser-94: don't change your shell
<unmatched-paren>change the shell your terminal launches instead
<unmatched-paren>if you change it, it will break gux's startup scripts
<PotentialUser-94>okay got it
<PotentialUser-94>okay, so now when i try to reconfigure unattended upgrades with the (gnu services admin) line added, it throws an error parsing the derivation
<PotentialUser-94>one sec i'll pastebin it
<PotentialUser-94>does that mean there is an issue with the service package?
<efraim>is it strange that my /var/guix/db/db.sqlite-wal is 2.6 GB ?
<antipode>efraim: I think nckx has said something about huge sqlite-wal before
<efraim>nckhexen: ^^ any recollection?
<antipode>Something about encountering huge sqlite-wal,
<antipode>maybe something about things not being flushed?
<sektor[m]>Odd; mine is only 8.6m.
<antipode>efraim: Are any "guix build whatever" or such running?
<nckhexen>Uh, just that it was huge, about 7 GiB IIRC (but it was definitely ‘several gigglebytes’).
<nckhexen>I never fixed it, just nuked from orbit, sorz.
<antipode>Myself, I don't have a -wal, until I do "guix whatever ...", then there is _temporarily_ one.
<efraim>nothings building ATM, I did recently clear ~500 GB from my store though
<nckhexen>The WAL is definitely not supposed to be that huge. It is indeed something where writes stop getting flushed to the actual DB, and you just get a ‘pending’ log that grows, presumably forever.
<nckhexen>antipode: Yes, that's the working normal case.
<nckhexen>I don't have one right now either.
<antipode>It looks like "guix shell" counts as a "guix build"-like thing for the purposes of '-wal exists': guix shell --rebuild-cache --pure coreutils -- du --hu /var/guix/db/db.sqlite-wal
<nckhexen>Sure. There's nothing magical about ‘build’. ‘gc’ would surely do the same.
<antipode>I found a reproducer of 'WAL grows'
<nckhexen>Kabouik: <I heard it's not a good idea> If this is about a recent conversation in #emacs: there was a lot of nonsense in that. If you're experiencing actual bugs, of course, that's different, but don't worry too much otherwise.
<antipode>nckxhexen: "guix shell" could have been written to do a 'exec' instead of fork+exec, breaking the store connection (and, if I guess correctly, trigger the guix daemon to flush the DB)
<antipode>Or, alternatively, still do a fork+exec but disconnect before invoking.
<nckhexen>Sounds like working around a fixable flaw in the daemon to me.
<efraim>vacuuming the db didn't shrink the wal
<antipode>I considered it more of a small optimisation of "guix shell" that could be done independently of its usefulness as a work-around.
<nckhexen>I know I'm being super-helpful here but: nope, can confirm, I tried that then. Maybe there's other stuff I'll remember trying as soon as someone mentions it ☺ I just know I didn't find an easy fix, and time ran out. Glad it's receiving some attention now.
<nckhexen>(re: vacuum.)
<antipode>-- to make 'guix shell [...] -- foo' tail call into 'foo' (one less process that occupies memory), and such that the daemon can free a little memory.
<nckhexen>OK, that's a good point.
<antipode>nckhexen: On the work-around: IIRC, the guix daemon associates some stuff not with connection but with processes. And processes are identified by their pid, which is kept intact by 'exec'. Hypothesis: maybe that's why the work-around didn't work.
<antipode>(Untested & unverified)
<efraim> has the information I've been looking for
<apteryx>lilyp: are there plans to enable native compilation for emacs-minimal? it'd be nice that it's enable out-of-the-box for all users.
<apteryx>then we wouldn't need the note added in 4cb5144df0
<efraim>`sudo -E sqlite3 /var/guix/db/db.sqlite 'PRAGMA main.wal_checkpoint' lets you know if everything in the WAL has been written to the database yet
<efraim>if numbers 2 and 3 are the same (ie 0|199|199) then 'PRAGMA main.wal_checkpoint(TRUNCATE)' is probably safe, it'll force a checkpoint if needed and lock the database, write everything from the wal into the database and zero the wal
<linj>I am preparing to install guix and find this issue: cryptsetup-static aborts opening LUKS2 volume with Argon2i PBKDF ( Hope it could be fixed.
<linj>Seems like what prevents that patch is the rebuilding of a whole system. Is that a valid argument against a patch?
<nckhexen>It's not an argument against the patch.
<antipode>many rebuilds -> crore-updates
<nckhexen>It means that it's expensive to test and to apply.
<antipode>* core-updates
<dgcampea>what do I need to do to get (invoke "sh" ...) working for a trivial-build-system package?
<dgcampea>I am getting invoke-error exit-status 127
<dgcampea>I assume it's not picking sh up?
<unmatched-paren>dgcampea: either that or the script doesn't exist
<unmatched-paren>you probably do need to add bash to the inputs
<unmatched-paren>native-inputs, sorry
<unmatched-paren>i doubt trivial-build-system has any implicit inputs...
<nckhexen>Yah there is nothing to ‘pick up’.
<nckhexen>Nor any mechanism (PATH) to do so.
<dgcampea>hmm... adding bash to native-inputs didn't fix it
<nckhexen>The ‘trivial’ in ‘trivial-build-system’ stands for ‘did you mean to write gnu instead friend’.
<nckhexen>See above.
<dgcampea>the script exists for sure
<nckhexen>If you want PATH, you get to make it from sticks other things you find in the wilderness.
<nckhexen>* &
<dgcampea>nckhexen: the package is not the typical autogen/autoconf :(
<nckhexen>Many g-b-s packages aren't.
<nckhexen>Anyway, setenv is your friend, but the t-b-s really isn't, watch your back.
<dgcampea>I can swap with trivial with gnu?
<dgcampea>trivial with gnu**
<nckhexen>It's a common thing to do. Not the swapping as such, but using gnu- and just deleting the phases you don't need/want.
<unmatched-paren>hmm, so, i'm making Yet Another home-emacs-service-type (though I intend to upstream this, making all the other existing home-emacs-service-types obsolete, hopefully...) and I get ``stdin is not a tty'' and a hang when I run emacsclient while it's running...
<nckhexen>gnu- still provides some other nices-to-have like shebang patching & stripping binaries & validating runpaths.
<unmatched-paren>Presumably this is because of shepherd's daemonisation, but rde's version doesn't have this problem.
<nckhexen>Seriously, trivial- is not meant as a good base for fun packaging.
<unmatched-paren>My home-emacs-service-type runs ``emacs --fg-daemon -q -l /gnu/store/...-init.el'', and their's does ``emacs --fg-daemon=...''.
<unmatched-paren>I don't see why theirs would work but mine wouldn't.
<unmatched-paren>I mean... I'm not really even sure theirs works, but I'm working off the assumption that it does :)
*mbakke has been manually building i686 stuff on berlin and the staging weather is now nicer than on 'master' ... I think the branch can be merged now.
*mbakke is a biological CI system
<mbakke>I should do some Cuirass hacking one of these days
<dgcampea>Other than '#:use-module (guix build-system cmake)', do I need anything else? I'm getting ERROR: In procedure %resolve-variable:
<dgcampea>Unbound variable: cmake-build
<dgcampea>the shell script was invoking cmake behind the scenes so I thought on using cmake directly
<unmatched-paren>dgcampea: You've set (build-system cmake-build-system), yes?
<dgcampea>unmatched-paren: yes
<dgcampea>(ignore the build-script variable, it was from a previous refactoring)
<unmatched-paren>Remove #:modules.
<unmatched-paren>And its value, of course.
<dgcampea>?? ok, it did something
<unmatched-paren>Because (guix build utils) is imported by default.
<dgcampea>why would it influence though?
<unmatched-paren>And because you overrided the modules without splicing %cmake-build-system-modules into the list, (guix build cmake-build-system) wouldn't have been imported, which is used by the default build phases.
<lilyp>apteryx: there's no point, different emacs, different cache
<dgcampea>unmatched-paren: I see
<dgcampea>does the cmake-build-system allow changing the directory before it runs cmake? I'm trying to map this shell snippet into the package definition (
<unmatched-paren>dgcampea: You can just add a phase to (chdir "DIR").
<Kabouik>nckhexen: Yes, that was related to that discussion in #emacs, but it was also enlightening because I was probably a bit shortsighted thinking I could enjoy all emacs powers from the terminal. I know emacs is more than a text editor, but I admit this is always how I was seeing myself using it, and in that sense I just wanted to compare it to vim or kakoune… In the terminal. Now that also brings other issues: I cannot focus emacs windows with the
<Kabouik>cursor (if using -nw), spawning frames instead of windows seems to be difficult because emacs -nw doesn't know how to start new terminal windows (except if I do it manually with a shell command), etc.
<Kabouik>But now with the GUI, I'm also facing issues. I already removed the tool/scroll/menu bars to make it more minimal, but haven't succeeded with transparency, and am still struggling with the concept of emacs windows splitting the frame instead of opening new actual WM windows that I could manage with Sway. Slowly learning.
<dgcampea>unmatched-paren: it does slightly more than a chdir
<dgcampea>it does a chdir but it wants to call cmake ..
<clemens3>always build emacs with --without-x
<clemens3>then can forget about -nw flag
<clemens3>xterm is my frame
<clemens3>aeh, wrong channel!
<Kabouik>Speaking of, I'm using emacs, and am not sure that emacs GUI does a very good match with Wayland. Every time I found resources about transparency and such, they were referring to X, not Wayland.
<unmatched-paren>Kabouik: Wayland support is very new.
<unmatched-paren>It only works with PGTK, for one...
<Kabouik>What is PGTK?
<unmatched-paren>It was a project to make Emacs run on purely GTK without custom rendering.
<unmatched-paren>It was merged into Emacs mainline fairly recently. Not even sure if it's enabled by default yet.
<unmatched-paren>Actually, I know it isn't, because on Guix, you need to use emacs-next-pgtk to get support for it.
<Kabouik>So I installed emacs-next earlier today (after trying emacs-minimal first, and then just emacs). Do I understand it right that emacs-next-gtk to get the best possible experience on my Wayland system?
<Kabouik>*is recommended
<unmatched-paren>It will just run with XWayland if you don't use PGTK.
<Kabouik>(This would really help if things like that were stated in the description of all those emacs packages; most of them share the same generic description and moggles like me who are undocummented cannot tell the difference)
<Kabouik>Thanks, settling with emacs-next-gtk then!
<lilyp>pgtk is currently just a configuration option for emacs-next, i.e. it will become mainline in emacs 29
<jonsger>mbakke: is it still worth to reconfigure on staging?
<jonsger>or is the merge already almost done