IRC channel logs

2020-11-27.log

back to list of logs

<nckx>OriansJ: Builds fine here but I can't reproduce that derivation name.
<nixo_>Hi guix! Is there an `alist-replace`-like function that replaces many items?
<mbakke>nixo_: (fold alist-replace ...) ? :P
<nckx>OriansJ: I get /gnu/store/ryad0k15xl0i33fdsbq7bidr45b83wna-mpg123-1.26.3.drv
<nixo_>mbakke: probably:D I'll try with it, thanks
<nixo_>yep it was that easy
***sturm is now known as Guest18924
<roptat>alright, I managed to build these packages: https://framagit.org/tyreunom/guix-android/-/blob/master/android/packages/android-tools.scm
<roptat>I'm very happy that these package definitions look so clean :)
<roptat>it hides the real horror story on the build side ^^'
<roptat>is there any video player that would let me set audio and video speed independently?
<ryanprior>You can install Guix System 1.2 on a Pinebook Pro?
<leoprikler>roptat: not directly, but you could try some outlandish ges pipelines
<leoprikler>[of course ffmpeg might work as well]
<ryanprior>roptat: you mean you want to split the audio stream off and play it out of sync with the video?
<guixy>trouble with gitolite. Shouldn't I be able to clone gitolite-admin with the key I put in the config?
<guixy>Do I need to specify myname@ip-address:gitolite-admin if the key is in myname.pub?
<guixy>Or should I keep it git@ip-address:gitolite-admin
<roptat>I ended up using the preview from openshot ^^
<roptat>separating the audio and video, and making the video channel a bit slower
<roptat>after a bit of trial and error, I found the right value
<leoprikler>apropos openshot, I've packaged pitivi a few days ago
<leoprikler>[on staging, not yet applied]
<ryanprior>Oh nice, I like pitivi and have been using the Flatpak for it.
<guixy>git ls-remote git@192.168.1.8:gitolite-admin
<guixy>git@192.168.1.8's password:
<guixy>I didn't set up a password for git on that machine. I don't see an option to do that.
<leoprikler>ryanprior: 44806, though perhaps it needs a bit of cleanup (especially #3 is cheeky)
<euandreh>How can I get the logs of a shepherd service? Is there a 'herd logs nginx' akin to 'journatctl -u nginx'?
<leoprikler>not really, but most should use syslog, i.e. log to somewhere in /var/log
<euandreh>I see logfiles there, but they're mostly empty. I though I could find more entries on the startup of a service when shepherd started it
<leoprikler>Shepherd's own logs are pretty thin in that regard
<leoprikler>if your application logs to stdout you may want to redirect that for debugging
<leoprikler>[i personally have an emacs service, that i can see in the command line logs, but not in the file logs]
<leoprikler>welp, time to sleep
<euandreh>ty leoprikler
<euandreh>good sleep
<xelxebar>When running shepherd as a normal user, I'm getting "incompatible bytecode version" for all compiled modules.
<xelxebar>Just did a pull, upgrade, reconfigure, and reboot yesterday, so naively I'd expect guile versions to be in alignment
<xelxebar>What am I missing?
<OriansJ>nckx: odd that we get different checksums for the same guix version; and when I attempt to look at the file, it appears the path is wrong (as in the parent folder doesn't even exist)
<save-lisp-or-die>any common lispers here who happen to know how to get the sdl2 common lisp system to find libsdl2 ? On other linux systems, cffi uses dlopen to find libsdl2 (i.e. I don't have to know), but this doesn't seem to be the case on guix. Any pointers?
<kozo[m]>save-lisp-or-die:
<kozo[m]>save-lisp-or-die: Yes, one sec. Let me get you the code needed
<kozo[m]>You need to have sdl2 installed locally with guix install sdl2
<kozo[m]>(ql:quickload :cffi)
<kozo[m]>and then (pushnew #P"/home/user/.guix-profile/lib/libSDL2-2.0.so.0"
<kozo[m]> cffi:*foreign-library-directories*
<kozo[m]> :test #'equal)
<kozo[m]>save-lisp-or-die: I am a fairly new lisper so if you know a cleaner way of doing it, please share after
<save-lisp-or-die>ah-hah. I was looking in /gnu/store/... for the right path
<save-lisp-or-die>and I have several profiles
<kozo[m]>I had tried that but never got it to work from there
<save-lisp-or-die>I just installed guix this morning and am just trying it out - didn't realize that $HOME/.guix-profile had shared libs too
<save-lisp-or-die>very cool.
<kozo[m]>They only show up there if the program is installed locally. When it's installed with config.scm, it won't show up there
<kozo[m]>They go to /gnu/store instead
<save-lisp-or-die>I *think* I've been installing everything "locally". I don't really know. Installing via `guix install`
<kozo[m]>When you install using "guix install" only that user will see the program. If you install a package using the config.scm, it will be available globally to all uers
<kozo[m]>users*
<save-lisp-or-die>thanks. I've been scanning the manual but haven't had the heart to slog through it with any diligence yet.
<kozo[m]>save-lisp-or-die: =]
<kozo[m]>For my emacs config, I do a (with-open-file (pretty-print.......)) to have it make my .emacs file anytime I run the config. Easy to move it between machines
<kozo[m]>Would you like the code for that?
<save-lisp-or-die>I'm not sure I follow
<kozo[m]>Do you use emacs for your cl programming?
<save-lisp-or-die>I do.
<kozo[m]>In my config.scm (your core scm file that says what happens on your system), if you have a complicated .emacs file you don't want to write out, you can make a define (defun) to have it write the file for you to a location.
<save-lisp-or-die>ahh - I'm not running the guix os, I've installed the package manager along side my systems package manager (debian)
<kozo[m]>Oh ok
<kozo[m]>Carry on then.
<save-lisp-or-die>thanks much
<save-lisp-or-die>I've been lurking all day :)
<kozo[m]>You're welcome
<procra>hi guix, i try to install guix on a KVM machine from the ISO because i couldn't figure out how use dvorak on the the VM image. and i am having some problems when the live iso start, i have a black screen and i don,t know what to do.
<txgvnn>/gnu/store/87kif0bpf0anwbsaw0jvg8fyciw4sz67-bash-5.0.16/bin/bash: -c: line 0: unexpected EOF while looking for matching ``'
<txgvnn>/gnu/store/87kif0bpf0anwbsaw0jvg8fyciw4sz67-bash-5.0.16/bin/bash: -c: line 1: syntax error: unexpected end of file
<txgvnn>
<txgvnn>/gnu/store/87kif0bpf0anwbsaw0jvg8fyciw4sz67-bash-5.0.16/bin/bash: [11:22:15]: command not found
<txgvnn>
<txgvnn>/gnu/store/87kif0bpf0anwbsaw0jvg8fyciw4sz67-bash-5.0.16/bin/bash: [17:11:05]: command not found
<txgvnn>
<txgvnn>Sorry everyone, my erc has problem with znc server, I still don't know to fix :(
<ece>So I've installed guix with luks encryped root, but when I boot into grub I can't enter my password xD, has anyone else had this issue?
<divoplade>You need to input it as if you were using a QWERTY keyboard
<divoplade>(only for the first time, after that input it normally)
<divoplade>Once you log in, run cryptsetup luksAddKey /dev/sdXX (where XX is the partition of root), type your password, switch to a QWERTY keyboard, and type it again twice so that there are 2 passwords to unlock your root: your password, and your password as typed on a QWERTY keyboard
<divoplade>And I need to learn to tag who I'm responding to, so that's ece now.
<vits-test>xelxebar: hi. try install guile-3.0-latest to ur user profile (if not yet).
<vits-test>
<divoplade>I'm not sure that answers the question, but that's the issue I got so I figure it can be relevant
<divoplade>Otherwise, please explain how you can't enter your password. Do you have a keyboard? Do you need assistive technology?
<lle-bout>nckx: so I tried, after libffi patch, powerpc64le-linux-gnu gcc-stripped-5-5 build of bootstrap tarballs isnt reproducible across Linux kernel versions (tried Debian Buster host and GNU Guix host)
<lle-bout>however, building on the same system repeatedly is reproducible
<lle-bout>tired now but will diffoscope soon
<wehlutyk>hello guix o/
<abcdw>wehlutyk, hi!
<wehlutyk>are any julia users having problems with Makie.jl? I'm trying to get the GLMakie backend to work and having issues, but not sure if it's a guix config problem or one of the makie/opengl/glfw issues
<civodul>Hello Guix!
<maav>good morning, civodul!
<mothacehe>hello civodul :)
<civodul>i sent a v1 of the ld.so.cache idea that rekado_ came up with the other day: https://issues.guix.gnu.org/44899
<civodul>how's everything, comrades?
<cbaines>I've finally succeeded in writing the script to submit builds from the comparison of two branches, I'd been procrastinating for days!
<cbaines>So I'm now building packages for staging https://data.guix-patches.cbaines.net/revision/24c86ad0c05bac966c3f5f3c5f40c6b0945adf09/builds
<civodul>yay!
<civodul>that should make mbakke and the rest of us happy
<maav>civodul: IMHO you should already push the second patch from #44899 ;-)
<civodul>ah right :-)
<maav>cbaines: looks awesome :-)
<civodul>but that's a full rebuild so i tend to batch them
<mbakke>cbaines: awesome :-)
<civodul>cbaines: BTW, glib failed (test timeouts) on guix-x15, even with -c1, but it succeeded on overdrive1
<mbakke>civodul: those patches look awesome, too!
<mbakke>I'm not sure where that glib derivation comes from
<mbakke>I mean, there are substitutes for gtk+ on armhf
<civodul>the replacement glib
<mbakke>aah
<cbaines>civodul, that's interesting that they managed to pass
<civodul>the ld.so.cache thing could be a welcome improvement
<civodul>cbaines: i think it's just that the overdrive is more beefy
<civodul>the timeouts in the test suite must be too short for the low-end ARMv7 boxes
<cbaines>anyway, in this specific case, I think that specific test seems to do more harm than good I think
<cbaines>as you're here mbakke, do you have any thoughts on what's needed before merging staging?
<cbaines>as above, I've got the patches instance of the Guix Build Coordinator building staging packages, which should reveal more failures if there are any
<mbakke>cbaines: I'm not aware of any regression (will check the data service later), so now I'm just waiting for the weather to improve
<cbaines>mbakke, cool, that's good to know
<civodul>ah ha, but will that ever happen? :-)
<civodul>mothacehe: your Cuirass "periodic job" trick seems to have a positive impact already
<civodul>it's not even 11AM and there's no GC running
<cbaines>I set out some thoughts on making staging/core-updates easier for substitute servers here https://lists.gnu.org/archive/html/guix-devel/2020-10/msg00401.html
<civodul>cbaines: we're only building the "core" subset for core-updates until it's frozen
<civodul>it's good to have that subset built continually because it makes it easier to have on the branch
<mbakke>*hack
<mbakke>for 'staging', I tend to batch up all the big changes (mesa, imagemagick) at once, but there is a lot of room for improvement there ... recently I also started building for x86_64 only outside the "freeze" cycles by changing the cuirass specification.
<cbaines>civodul, yeah, I know that was fixed recently
<mbakke>civodul: I suppose the ld.so.cache thing will fix https://issues.guix.gnu.org/26048 :-)
<maav>civodul: regarding the third patch from #44899, is there a reason to use /tmp instead of a relative path?
<civodul>mbakke: oh right, i had forgotten about that failed experiment
<mbakke>Oh, my MUA hadn't noticed that 26048 was actually closed "wontfix".
<civodul>maav: it's a throw-away ld.so.conf and we know /tmp exists in the build environment so that seemed right
<maav>civodul: i'm afraid of different builds getting the same file, although the fs isn't be shared, is it in any case?
<zzappie>h
<leoprikler>hello, anyone here using inkscape?
<leoprikler>trying to add emoji in text leads to mojibake and i have no idea why
<wehlutyk>a quick question on python packages: when a python library uses tox, should the guix package use tox directly or use the commands inside the tox.ini file?
<db48x>leoprikler: a google search suggests that you're not the only one: https://inkscape.org/forums/questions/calling-all-users-emojissymbols-in-inkscape/
<civodul>maav: ah no, there's no risk... unless you use --disable-chroot i suppose
<civodul>hmm
<civodul>maybe that's a good reason to change it
<civodul>yeah
<civodul>storing it under $NIX_BUILD_TOP would solve that
<leoprikler>db48x: so preloading fontconfig should do the trick?
<db48x>no idea
<leoprikler>nope, still cursed
<db48x>I tried it myself (in Fedora, not Guix), and instead of mojibake the emoji just didn't show up in the document at all (but they do show up in the text properties window)
<db48x>the characters are correct in the svg file too
<leoprikler>hmm perhaps I could do the rendering with cairo
<db48x>what are you trying to render?
<leoprikler>an svg, that contains emoji, duh
<db48x>ah
<db48x>or a web browser
<leoprikler>eog renders it partly correct
<leoprikler>inkscape has some custom properties, that won't work obviously
<OriansJ>guix 1.2.0 fails to successfully complete a single guix pull without substitutes.
<OriansJ>on Debian 10.6.0 AMD64 (default install, single partition, no swap)
<luhux>If you can, please use pastebin to post more information
<civodul>OriansJ: yes, or if you have enough details on how to reproduce, it'd be great if you could post it to bug-guix@gnu.org
<luhux>For example: screen output, or related logs under /var/log
<luhux>Is there any way to reduce the RAM usage of guix environment --container?
<luhux>I wrote my programming environment in mainfest.scm and used guix environment --container ... to start it, but I found that it takes up too much RAM
<civodul>luhux: i'm not aware of memory usage issues here, do you have a reproducer?
<luhux>This seems difficult...
<nckx>OriansJ: Your /gnu/store/r5qflfi674s336pbaqkphiadqyj2psq0-pulseaudio-13.0/lib/pulseaudio/libpulsecommon-13.0.la does exist on my system and is a valid-looking libtool archive.
<luhux>Using the htop tool, there is no way to see that a certain process takes up too much RAM, but the memory will be freed after closing the container
<chrislck>Ok I feel silly for asking for help - if I want to test build an updated guix package how do I (1) copy from guix search <package> to local ~/package.scm (2) build using ~/package.scm
<luhux>chrislck: Maybe you can use git with guix time-machine
<chrislck>oops not quite
<chrislck>Ok I feel silly for asking for help - if I want to test build an updated guix package how do I (1) copy from guix edit <package> to local ~/package.scm (2) build using ~/package.scm
<civodul>chrislck: first, when you copy the package file, you must change the module name in the 'define-module' form at the top so that it matches the new file name: https://guix.gnu.org/manual/en/html_node/Package-Modules.html
<civodul>then you can change the file so that its last expression is the identifier of the package
<civodul>at that point you can do "guix build -f package.scm"
<civodul>does that make sense?
<chrislck>identifier = ?
<nckx>(define-public <identifier> <package>)
<nckx>So: (define-public foo (package ...)) <newline> foo
<chrislck>ok
<chrislck>ok building...
<chrislck>this package file has nothing to do with the manifest.scm that I've seen on mailing list?
<chrislck>(in case anyone's asking, I'll aim to update gnucash package)
<chrislck>btw thanks for the hand holding :)
<nckx>chrislck: Not really related to manifest beyond ‘they both deal with packages’.
<nckx>A manifest is a list of packages you want.
<chrislck>ok
<chrislck>building \o/
<chrislck>where do I send the patch to
<nckx>guix-patches at gnu dot org.
<nckx>After reading the ‘Contributing’ section of the manual.
<nckx>& thanks!
<pineapples>o/
<nckx>\o
<soheil>Hello friends,
<soheil>What is the best way to share files from Guix to Windows in Boxes?
<soheil>(Windows is in GNOME Boxes.)
<soheil>never mind 😴️
<chrislck>so: what exactly is the benefit of maintaining a gnucash-manifest.scm in the project root? does it mean I could get a fresh Linux machine, install guix as foreign distro, cd gnucash, do some guix magic and be able to build gnucash?
<pinoaffe>hi guix, I'm trying to `guix system init` a system with efi-grub-bootloader to an external hard drive from a non-uefi guix system, and I'm getting errors along the lines of issues.guix.info/28768 - is this to be expected? is there some way to circumvent this or do I need to boot in uefi mode first?
<mbakke>pinoaffe: it is not possible to use grub-efi on a non-EFI system (this limitation is not Guix-specific)
<pinoaffe>mbakke: right, thanks, I thought I vaguely recalled something along those lines
<OriansJ>civodul: sure; here are the steps I performed prior to the guix pull: https://paste.debian.net/1174587/ using this hardware: https://paste.debian.net/1174593/ on debian 10.6.0 (only install standard-utils and nothing else); all in a single ext4 partition with no swap.
<pineapples>Question: when replying to an existing bug report, for example 111, do I have to add "bug#111 [the title of the bug report]" when sending an email to 111@debbugs.gnu.org; or may the e-mail subject be empty?
<pinoaffe>so I guess that means that you can't switch from non-efi to grub-efi using `guix system reconfigure` either, right?
<mbakke>pinoaffe: indeed, the system must be booted in "EFI mode" for GRUB to work
<mbakke>pinoaffe: the reason is that grub-efi needs to update EFI variables on the mainboard(!)
<OriansJ>I can provide exact bios information and more detailed steps for installing debian if so required
<mbakke>OriansJ: what error are you getting?
<pinoaffe>mbakke: yup, tho I guess that those EFI variable updates wouldn't be of use to me regardless since I intend to boot the disk in another system anyhow
<OriansJ>gnutls-3-16 fails to build when one does guix pull without substitutes
<mbakke>pinoaffe: indeed, it's a strange design ... now there is a way to create a "standalone" GRUB that works on any EFI system (it's how the installer works), but that functionality is not currently exposed as a bootloader option (would be nice tho)
<OriansJ>correction: /gnu/store/qm3l79ic89qpjjd8avqxd81425v4wvv5-gnutls-3.6
<civodul>OriansJ: re GnuTLS, could it be: https://issues.guix.gnu.org/44559 ?
<civodul>cbaines: hey BTW, wasn't it going to be Patch Review Day? :-)
<divoplade>Hello :) I'm replacing cuirass with a server-side update hook that builds all new commits and attaches the outcome as a git note to the commits, or rejects the push. It works, but it's a little difficult to extract CI outcomes.
<divoplade>Do you know if cgit can serve notes as blobs?
<euandreh>divoplade: Good question, I have never seen it on cgit serving git notes
<euandreh>You're using the git notes to see the CI reports?
<divoplade>To store the outcome of the CI
<divoplade>The reports are displayed during git push
<euandreh>got it
<euandreh>BTW, do you mind sharing your update-hook? I'm now interested 😬
<cbaines>civodul, I haven't done anything about that yet, I'm busy with other things at work.
<muthuh>Hey, 'gsettings' went missing on me, which package provides it would you know? On Arch it's glib2 - no such pkg in guix. Many thanks.
<chrislck>ok how best to run my newly-built package?
<Formbi>muthuh: glib:bin, I think
<chrislck>oh cool guix envionrment -l mypackage.scm
<euandreh>chrislck: yep, I was about to double check that on the help to say it
<roptat>that will not let you run the package I think, it will drop you in an environment with all its dependencies instead
<muthuh>Formbi: Nice one I'll try that.
<chrislck>roptat: correct... now where do I find the build package?
<euandreh>hmm, that's true. Maybe 'guix environment --ad-hoc -l package.scm'?
<euandreh>I don't recall if that works, though, that's just me guessing
<roptat>not sure...
<chrislck>seems to work
<roptat>oh, good
<euandreh>^_^
*civodul digs the RUNPATH rabbit hole
<chrislck>annoying that --ad-hoc -l package.scm doesn't load the prebuilt package
<divoplade>euandreh, you need to create a file named "ci.scm" at the root of your project, that evaluates to something to build (for instance, a dist-package) with the recursive local-file (so... directory) named "source/" as the source. Then you can use this as an update hook: https://paste.debian.net/1174609/
<euandreh>👀
<divoplade>You may need to tweak what you put in the note, this script works for a dist-package because it outputs only one .tar.gz file
<divoplade>the refs/notes/dist is arbitrary, you can put whatever you want inside refs/notes/*
<euandreh>ack
<divoplade>If you put refs/notes/commit, and you have a message as a note (and not a blob), then it will appear in git log
<divoplade>(provided that you pulled the notes from the server)
<euandreh>your mean refs/notes/<sha>?
<divoplade>No, that's sort of the name of a branch
<divoplade>If you have a CI that outputs a "dist", you can use refs/notes/dist
<divoplade>If it outputs a "thing", you would use refs/notes/thing
<euandreh>hmm, got it
<divoplade>That's not very important, but it avoids conflict with other uses of notes
<roptat>is there a way to know which architecture I'm building on and the target from the build side?
<roptat>I'm told %current-system doesn't exist on the build side
<save-lisp-or-die>hi #guix ! What advice to you have to offer for conserving disk space? I've installed the guix package manager along side my debian install and I have a somewhat small disk, and am curious about what guix users have to say about keeping a tidy filesystem.
<roptat>save-lisp-or-die, you can run "guix gc" to collect package that aren't used by any generation of any profile
<roptat>remove some old generations you won't use anymore to collect more garbage :)
<chrislck>Some errors :( https://pastebin.com/raw/eM07BWyb gtg for the night now
<save-lisp-or-die>roptat thanks much
<chrislck>otherwise it does actually run, surprisingly
<chrislck>and my, this build is super speedy
***txgvnn is now known as Guest77336
<guix-vits>sneek: seen anyone?
<sneek>Nope.
<guix-vits>s'neek the lair, aha.
***Guest77336 is now known as txgvnn`
***txgvnn` is now known as txgvnn
<roptat>found it, there's actually a keyword argument to the phases, called system (and target)
<zzappie>wat
***sneek_ is now known as sneek
<zzappie>sneek: botsnack
<sneek>:)
<zzappie>Hey guix! Is anyone runing guix on odroid XU4?
<zzappie>I know C2 was exclueded because it needs binary blobs to boot. How do I check wheter it is the case with a board?
*nckx genuinely surprised that %default-locale-definitions doesn't contain ‘eo’.
<roptat>mh... guile is sometimes a bit cryptic... Undefined symbol #f
<roptat>how is #f not defined? ^^'
<civodul>#f is not a symbol
<civodul>how d'you get that error?
<civodul>perhaps "Undefined symbol" actually comes from some library that you're using?
<roptat>when building a source file
<roptat>I suppose it's my macro doing something wrong
<roptat>how do I debug a macro again?
<civodul>hmm, "debug a macro", what do you mean? :-)
<civodul>you can use ,expand at the REPL to see the result of macro expansion
<civodul>(or ,optimize)
<civodul>but that's about it
<roptat>thanks
<civodul>there's no macro stepper like Racket has
<civodul>you only get to see the final macro expansion result
<civodul>(C-c RET e in Geiser)
<roptat>oh, it's actually when defining a function
<nckx>My SDDM+Sway desktop is missing $LANG everwhere, from the DM down. It's present in /etc/environment and set correctly on VT logins. Who's responsibility is it in Wayland?
<roptat>I'm trying to define this: (define* (generate-genrule-makefile config #:key #:allow-other-keys) ...)
<roptat>why is that wrong?
<euandreh>roptat: Isn't #:key missing a name after it?
<euandreh>like '(define* (generate-genrule-makefile config #:key a-key #:allow-other-keys) ...)'
<roptat>yay... I thought it would allow be to say "accept any keyword argument" but it seems you need at least one
<euandreh>"accept any keyword argument" is the #:allow-other-keys part
<roptat>yay, but it doesn't work when it's alone
<roptat>mh... when specifying --target=i586-linux-gnu, the target keyword in the phases is still set to #f (I made sure to not have #:target as a private argument)
<euandreh>what do you mean?
<roptat>(define* (test #:allow-other-keys) 1) won't work
<roptat>but actually, I can (define* (generate-genrule-makefile config . _) ...) and it works
<roptat>that shouldn't be -linux- in that target, oops
<euandreh>yeah, I was about to suggest using (... . _ignored) too
<euandreh>also (... #:rest ignored) is the same as an alternative
<euandreh>divoplade: this idea of attaching CI logs to git notes is intriguing, I'm liking it. Have you seen it done somewhere else?
<divoplade>Unfortunately, no
<divoplade>If you want to run the CI after the push is accepted (like what we see on gitlab, github), you might use another hook than the update hook
<guix-vits>nckx: hello. is environment working w/o sddm as in `sway` from tty? i've LANG set. i haven't updated in a month, tho.
<nckx>guix-vits: Yes, that's what I mean by VT. Sessions spawned by a getty have $LANG just fine. Interestingly, it seems that sessions spawned by SDDM do have it... if I log in manually (I prefer auto-login but managed to kill Sway in an unrelated kerfuffle).
<nckx>So that needs some further confirmation/investigation when I can afford to kill my desktop again. Trying to launch another SDDM didn't end well.
<guix-vits>nckx: do u use sddm only for testing?
<Gooberpatrol66>What can cause this type of error?
<Gooberpatrol66>nathan@guixlaptop ~/guix $ guix system disk-image /home/nathan/guix/config-guixrig.scm
<Gooberpatrol66>guix system: error: /home/nathan/guix/config-guixrig.scm:129:12: no value specified for service of type 'xmr-stak-rx'
<nckx>guix-vits: No, I've always used it.
<nckx>Gooberpatrol66: You're using (service xmr-stak-rx-service-type) while Guix expects (service xmr-stak-rx-service-type VALUE), i.e. xmr-stak-rx needs to define a default value.
<Gooberpatrol66>nckx: what does the default value contain?
<nckx>Gooberpatrol66: See commit 97e98e22691d3d1ecf35afc2ba2c961fdbd3cc7e.
<nckx>I'd link to https://git.savannah.gnu.org/cgit/guix.git/...commit but it's down.
<nckx>It's pretty trivial if your (xmr-stak-rx-configuration) constructor returns sane defaults.
<nckx>This being a custom service I can't say.
<Gooberpatrol66>nckx: what if there's nothing I care about configuring for the service? Can I make it not take a configuration?
<nckx>Sure, it can be '().
<nckx>Savannah don't feel so good. I assume guix pull is broken at the moment but I'll away the first ‘is guix pull broken?’ here before adding my traditional klaxons to the topic.
<nckx>*await
<guix-vits>nckx: in w3m i get "certificate is not (yet) valid"
<guix-vits>sorry for noise, it's ntp
<nckx>guix-vits: The issue is that the machine's RAM and swap are almost full, so that sounds like a different issue.
<nckx>OK.
<mbakke>huh, I just noticed alacritty sets LD_LIBRARY_PATH for ... well, everything. That does not seem right..
<nckx>‘hmm, we may or may not be under attack - i'm seeing a drastic increase in number of nginx connections’ ah great here we go again.
<nckx>mbakke: Er, no. But it sounds familiar.
<nckx>Weird, no bug.
<roptat>nckx: getting a 504 from savannah after a long wait when trying to guix pull
<nckx>Right. Here come the klaxons.
<guix-vits>also knowing that my default date is in 2013, it was not clever to set up a https "constraint" site.
<guix-vits>* choose a
***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | ⚠️ ‘guix pull’ servers are downish ⚠️ get Guix at https://guix.gnu.org | videos: https://guix.gnu.org/blog/tags/talks/ | bugs & patches: https://issues.guix.gnu.org | paste: https://paste.debian.net | Guix in high-performance computing: https://hpc.guix.info | This channel's logged: https://logs.guix.gnu.org | Guix Days, Nov. 16-22: https://guix.gnu.org/en/blog/202'
***nckx changes topic to 'GNU Guix | ⚠️ ‘guix pull’ servers are downish ⚠️ get Guix at https://guix.gnu.org | videos: https://guix.gnu.org/blog/tags/talks/ | bugs & patches: https://issues.guix.gnu.org | paste: https://paste.debian.net | Guix in high-performance computing: https://hpc.guix.info | This channel's logged: https://logs.guix.gnu.org'
***ChanServ sets mode: -o nckx
<Gooberpatrol66>nckx: where do I put the '()? Inside the definition of xmr-stak-rx-service-type?
<nckx>guix-vits: The constraint should serve only as a sanity check of the date returned by your NTP pool. This sounds more like an ’allow large adjustment’ issue.
<nckx>Gooberpatrol66: (default-value '()) or so, yes.
<guix-vits>nckx: idk, large-adj is #t. openssh seem to be picky for me.
<Gooberpatrol66>nckx: I got a really huge backtrace when I tried to do that
<bdju>I'm getting a 504- oh I see the topic
<nckx>:)
<nckx>I know it's a bit subtle.
<guix-vits> /topic серверы загрузкамана кирдык
<nckx>Не кирдык, они просто отдыхают. 🦜
*nckx probably butchered that, sorry.
<nckx>‘oh, looks like someone requested the entire www-ja.git repo be archived [via gitweb], and that's hogging the server’
***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | get Guix at https://guix.gnu.org | videos: https://guix.gnu.org/blog/tags/talks/ | bugs & patches: https://issues.guix.gnu.org | paste: https://paste.debian.net | Guix in high-performance computing: https://hpc.guix.info | This channel's logged: https://logs.guix.gnu.org'
***ChanServ sets mode: -o nckx
<nckx>And we're back, thanks to bandali. <- all hail them
<bandali>happy to help ^_^
<bdju>yay
*save-lisp-or-die hails bandali
*bandali tips hat
<roptat>when are build tools added to a package's input?
<roptat>especially, when it doesn't use the gnu build system, how do they still get in the set of inputs? (like tar, patch, ... even gcc and co)
<cbaines>roptat, most build systems inherit from the gnu-build-system, so it still applies
<cbaines>I think there's some stuff related to that happening here https://git.savannah.gnu.org/cgit/guix.git/tree/guix/build-system/gnu.scm#n268
<luis-felipe>leoprikler: Speaking of Inkscape, is it just me or the Inkscape window got translucent?
<sneek>Welcome back luis-felipe, you have 1 message!
<sneek>luis-felipe, vits-test says: i imagined a shirt with the l. notice header before actual lyrics. better safe than sorry :|
<roptat>ah right, through standard-packages
<roptat>that is just the same as %final-inputs from commencement.scm
<leoprikler>luis-felipe: not really, but you can get gnome to glitch out windows arbitrarily
<luis-felipe>Hmm, but it seems to be Inkscape only, and I haven't touched GNOME settings...
<leoprikler>is it inkscape always or just that window?
<luis-felipe>Nah, it was something transient. It's ok now after restarting Inkscape.
<guixy>I'm trying to document what is required to use gitolite on guix.
<guixy>I got it working, then fiddled with something, and now it's not working anymore.
<roptat>alright, cross-compiling for arm-android :)
<guixy>It should be as simple as including the gitolite service as described in the manual, and copying/pasting the administrator public ssh key in the designated configuration field.
<guixy>But when I try to ssh as described (ssh git@server help) it prompts for a password)
*guix-vits hails bandali
<roptat>guix-vits, if that helps, I have these: https://framagit.org/tyreunom/system-configuration/-/blob/master/systems/ene.scm#L171
*roptat is wondering why building the bootstrap binaries is needed, when I have only native packages in the bag
<guix-vits>nckx: it's good
<roptat>I'm using clang which should already be able to cross-compile
<ss2>heya, has anyone got some recommendatins on video workshops on using guile with geiser in general? I'd really like to get into it a bit further. Reading and doing tutorials here and there haven't been enough for me yet.
<luis-felipe>I don't know of any, but it would be nice to have a Guile+Geiser video series.
<guix-vits>ss2: rule #1: run geiser in separate emacs, as it can freeze. THE END :)
<nly>news spreading faster than california wildfires: https://www.talospace.com/2020/11/guix-port-to-openpower.html
<pineapples>nly: Oh nice
<guix-vits>+1
<bdju>looks like the alacritty package has fallen quite a bit behind. it's on version 0.4.1 and 0.6.0 is now out
<nckx>It's a pain to update because Rust is a pain. I started to, then noticed the memory usage and went back to termite 😛
<pineapples>I tried to update the package declaration, but that failed miserably. Somebody else got to do it..
<nckx>Hm, I should have said something more motivating. It's more fun™ to update than other boring packages!
<guix-vits>... so much that packagers scar the table after every update: | | | | ...
<apteryx>it'd be useful to have an easy way to print out the full versions listing of a package, to share with upstream
<apteryx>of the dependencies of said package
<apteryx>hmm, apparently 'guix show package' contains the versions... nevermind :-)
<nckx>guix gc --references?
<nckx>Ah.
<apteryx>I thought it didn't for some reason.
<Formbi>I use st
<bdju>nckx: I stopped using termite mainly because it uses pango, and pango broke bitmap support
<Formbi>with the package transformation options installing the personal variant is a one-liner
<bdju>nckx: by the way, what part of the process of updating the package involves high memory usage? is it something like having to test it on your PC before you submit it for the build server?
<bdju>I have this annoying paste bug that I'm pretty sure is with alacritty and it may or may not be fixed in newer versions. occasionally my paste happens a bunch of times in a row instead of once.
<nckx>bdju: I meant during my regular use, not while updating. It wasn't prohibitive and shouldn't keep anyone from updating it (unlike unthings like Chromium that need gigs & gigs just to link).
<bdju>ahhh okay
<nckx>I'll install it & see.
<nckx>Maybe it was a bug.
<bdju>I did see in one of the changelogs a reduction in memory usage by 30%
<mbakke>speaking of alacritty, /me pushes a fix to remove the LD_LIBRARY_PATH wrapper
<bdju>ah yeah, the latest release reduced memory usage
<nckx>Oh yeah.
<guix-vits>i've a question on guile 3.0.4, will try ask on #guile tommorow: https://paste.debian.net/1174640/
<nckx>Wowsies.
<nckx>It uses marginally less than termite now.
<euandreh>divoplade: 'git notes --ref=refs/notes/ci show' hmmmmm, now I have the exact same question as you do: does cgit support showing notes?
<nckx>Damn, now I have to switch & update a Rust package :-/
<bdju>hahaha
<mbakke>it's terrible that it's not possible to centrally patch libraries with cargo-build-system
<euandreh>divoplate: TBH I was already running CI jobs on git's update-hook, but storing the logs on the server. Adding them to git notes is betterA
<mbakke>I had to patch rust-wayland-sys etc, in the alacritty definition
<mbakke>ooh, thanks for the gstreamer fixes on staging nckx
<mbakke>I'm gonna merge it tomorrow, probably
<nckx>False alarm. Memory usage of a new terminal is ~38M (termite ~39M). Run find /gnu/store in both and termite is using 200% CPU for termite's 70%, and currently using 1.4GiB. Now perhaps I need to tweak a scrollback setting (but I want 1000s of lines!) but, here's the kicker, allacritty's much slower, the end, bye.
<bdju>oh noooooooooo
<nckx>I'm sure alacritty's the bomb on a quad-GPU Ryzen & all. Good for them. Not on my Thinkpad.
<bdju>I'm just using a ThinkPad T440p on my end (although possibly a newer model than what you've got)
<bdju>notably without the nvidia graphics some models have
<nckx>Ehm, wtf. Alacritty is shows a whopping 100 lines of scrollback and is currently using 12 GiB of RAM.
<mbakke>uh
<bdju>well, that sounds like a bug
<nckx>I think so. That 100 is an honest guess because I selected the text to pipe through wc -l and now it's frozen.
<bdju>is this on the currently packaged version or a newer version, btw?
<nckx>Guix 828416e25f75be1cd0fc22d4939d81c0a4d754ec.
<nckx>Oh, right, that's my commit. Base is 3c881facced4cad373b7e0770fff9e5c5b01333f.
<bdju>not actually quite sure what to do with that hash or how to check my own
<nckx>guix pull --commit=<hash> or something. But it's basically current master, at most a day old.
<nckx>‘Current alacritty’ quoi.
<bdju>and that's the version of guix or alacritty?
<nckx>Guix. Alacritty is /gnu/store/di2j4xjsk8lrq5vv1wkkqy8y2hm6kixz-alacritty-0.4.1.
<nckx>-the trailing . of habit.
<nckx>bdju: Mine's an x230t with 3rd-gen Intel grafix, no external GPU.
<nckx>mbakke: Thanks for killing the LD_LIBRARY_PATH monstrosity.
<mbakke>nckx: np! it's the last thing you want set in your terminal of all places :P
<nckx>Imaging debugging something caused by that.
<nckx>*e
<mbakke>shivers
<mbakke>I'm gonna try that rust importer and update it this weekend if I find time...
<mbakke>but now, weekend!
<nckx>Party hard.
<nckx>Home. Alone.
<roptat>no... armv7a-linux-androideabi-ld: cannot find crtbegin_so.o: No such file or directory
<roptat>:(
<civodul>roptat: you've packaged Bionic?
<apteryx>oh, connection timeout to ci.guix.gnu.org
<apteryx>substitute: guix substitute: warning: ci.guix.gnu.org: connection failed: Connection timed out
<apteryx>is the ci under very high load?
<nckx>No, not at all.
<nckx>4.
<roptat>civodul, no I'm just trying to use the existing infrastructure
<roptat>also, I'm trying to use clang as a cross-compiler, so I'm not sure why guix wants to build croos-gcc
<civodul>--target pulls in the cross toolchain, which is the GNU toolchain by default
<roptat>I even made sure to remove gcc, glibc, binutils and ld-wrapper from the inputs
<roptat>ah, I thought it was just passed as an argument to the build system?
<guixy>I think I fixed my problem with gitolite by forcing it not to take passwords for ssh...
<civodul>roptat: the toolchain is among the implicit inputs that the build system pulls in
<civodul>same for the cross-toolchain
<roptat>I checked with guix graph -t bag, and my package doesn't pull in gcc etc
<roptat>I'm using my own build system, and instead of (standard-packages), I basically use (filter ... (standard-packages)) to remove gcc, glibc, binutils, ld-wrapper
***jonsger1 is now known as jonsger
<roptat>it seems that guix is trying to build the toolchain even before creating a derivation for my package
<apteryx>nckx: thanks. Weird. I'm still experiencing the failure.
<nckx>Any type of request? I'll blast it with a guix weather from here.
<apteryx>just trying to get substitutes
<apteryx>from a 'guix build'
<apteryx>now I got: substitute: guix substitute: error: TLS error in procedure 'handshake': The TLS connection was non-properly terminated.
<guixy>and it broke again after a reboot...
<civodul>apteryx: weird, it works for me; could you try to wget something from there?
<civodul>like: wget -O - https://ci.guix.gnu.org/n8awazyldv9hbzb7pjcw76hiifmvrpvd.narinfo
<apteryx>that worked
<civodul>ok
<civodul>maybe try with the URL for which you got the error above?
<roptat>I get a lot of these: "guix offload: warning: The 'system' field is deprecated, please use 'systems' instead."
<roptat>what does it mean'
<civodul>what do you mean "what does it mean"? :-)
<apteryx>the build-machine record now uses a 'systems' list of systems rather than 'system'
<civodul>see https://guix.gnu.org/manual/en/html_node/Daemon-Offload-Setup.html
<apteryx>the non-list flavor is still accepted but deprecated
<roptat>ah
<roptat>well, I meant "where does it come from" and where is this system 'system' field :)
<apteryx>ah, the connectivity error exists only from my offload machine
<apteryx>probably a network issue there
<roptat>the message could be better
<civodul>ah right, it lacks location info
<roptat>what about "please use 'systems' in your `machines.scm' instead."?
<civodul>or just add location info, no?
<roptat>sure
<apteryx>interesting that I get to see 'substitute: updating substitutes from 'https://ci.guix.gnu.org'... 100.0%' before 'substitute: guix substitute: warning: ci.guix.gnu.org: connection failed: Connection timed out'
<roptat>I don't know how to do that though
<civodul>roptat: https://web.fdn.fr/~lcourtes/pastebin/build-machine-location.html
<civodul>(untested)
<roptat>wait, what's the difference between host-inputs and build-inputs in a bag?
<roptat>thanks, I'll try that :)
*guix-vits ... probably "signals" is my issue.