IRC channel logs


back to list of logs

<nobodynogroup>hi :) I just installed guix. I hate to join and then immediately ask a question, but I've been trying for a few hours now, so has anyone got exwm working? I'm just not able to get it to show up in my xsessions on the default gdm. I started with xfce and openbox, and those two installed fine and are in my options, but even tho I have installed emacs-exwm, exwm does not show up in the list of WMs to choose from. Normally I would just add
<nobodynogroup>an xsessions .desktop file, but there isnt an xsessions folder in etc, and it looks like everything has .desktop files in the store directory, but so does exwm and it still doesnt show up
<nobodynogroup>I guess I'm mostly confused about how gdm is getting xsessions
<lfam>No worries about joining and immediately asking a question, welcome nobodynogroup
<lfam>I don't have an answer but I'm sure that somebody will chime in
*lfam brb
<nobodynogroup>thanks :) guix is great, very new to me, but I'm looking forward to learning all the stuff.
<Blackbeard>nobodynogroup: can you share your confg.scm
<jsoo>hey guix, how do i sneek a message later again?
<lfam>sneek: later tell jsoo: Like this
<sneek>Got it.
<jsoo>thank you!
<sneek>jsoo, you have 1 message!
<sneek>jsoo, lfam says: Like this
<lfam>sneek: later ask jsoo: Got it?
<sneek>Got it.
<lfam>I'm not sure how many verbs it knows
<pkill9>good evening guix
<jsoo>sneek: later tell cbaines: thanks for your patience with me. i'm working on purescript again.
<sneek>Welcome back jsoo, you have 1 message!
<sneek>jsoo, lfam says: Got it?
<jsoo>lfam: got it!
<jsoo>hi pkill9
<nobodynogroup>Blackbeard: i think i found the problem maybe, exwm.desktop is not in /run/current-system/profile/share/xsessions/ but I'm not sure how to get it in there correctly.
<seasharp>nobodynogroup: Ah yes, the xsessions folder. You're definitely on the right track, but I only know the Fedora method for adding gdm xsession files, so I can't be much more help than that
<mroh>try putting exwm in the system profile instead of the user one
<nobodynogroup>ah that sounds like a good idea, I will try that
<Blackbeard>nobodynogroup: yes, you should put all your WM or DE under config.scm
<nobodynogroup>hooray! that seems to have worked, the desktop file is in run now at least, thanks Blackbeard and seasharp
<nobodynogroup>and mroh
<anadon>Good evening all, would someone be able to help me fix a package I'm working on?
<seasharp>nobodynogroup: yw!
<seasharp>anadon: o/ hello, I'm terribly new at lisp and guix otherwise I'd help. I'm learning just from reading your package def tbqh
<anadon>It has some build failures due to not being able to find some binaries in PATH, but I'm not sure what I'm doing wrong.
<seasharp>Ooh someone suggested something to me recently that would help with that.....
<seasharp>I think it was something-replace-shebangs, in the build system?
<seasharp>cause I myself was having issues with loacting binaries during my package builds
<anadon>That sounds familiar, but I'm not sure what to do with it.
<seasharp>Here's what I'm looking at:
<seasharp>it's called patch-source-shebangs
<seasharp>there's guix docs somewhere one sec....
<seasharp>Patch shebangs encountered in source files so they refer to the right store file names. For instance, this changes #!/bin/sh to #!/gnu/store/…-bash-4.3/bin/sh.
<seasharp>according to...., but that's the best I can find so far
<seasharp>Oh, there's also patch-shebangs, which patches the installed executables, so you can modify inputs and outputs I suppose
<mbakke>lfam: great, thanks for testing. I'll make the merge tomorrow then. :-)
<mbakke>raghavgururajan: great, thanks! was worried that my intensive review might put you off the whole thing :-/
<raghavgururajan>mbakke Nah! Btw, I can send new patch-set in few hours. Would you be able to review and push it?
<mbakke>raghavgururajan: I'm going to bed soon, but can probably have a look over the weekend.
<raghavgururajan>mbakke: Cool! Thank you!
***familia_ is now known as familia
<lfam>Using inkscape from core-updates, the text in my inkscape files (svg) is all missing
<lfam>It seems like something changed regarding looking up fonts from the host system (Debian). I did run `fc-cache -rv`
<kolyad>On reboot after guix init, I am unable to boot as grub tries searches for a filesystem mapped to by luks, and not the luks device itself.
<lfam>sneek: later tell mbakke: After a while on core-updates on my workstation, I realized that fonts are broken for Guix packages on my foreign distro (Debian) after the core-updates merge. <>
<sneek>Got it.
<PotentialUser-28>I can't view videos (webm and mp4) in GNU Icecat.
<lfam>PotentialUser-28: Can you share the link of a video that fails to play?
<lispmacs>anyone here happen to own a TI-92+ calculator
<lfam>No, I have my TI-89 from 20 years ago, however
<PotentialUser-14>Hi, I want to install GuixSD with full-disk encryption on my librebooted x200, but I don’t want to enter my LUKS passphrase twice. On Parabola I put a key file inside the initramfs, but since GuixSD doesn’t use mkinitcpio or anything similar I don’t know how to solve my problem. Do you know of a secure (no key file in the store) workaround on GuixS
<Blackbeard>PotentialUser-14: hi
<Blackbeard>PotentialUser-14: this is usually a bad time because most developers are sleeping
<PotentialUser-14>ok thanks.
<anadon>I just can't figure out how to put down this warning! `patch-shebang: ./<some file>.py: warning: no binary for interpreter `python' found in $PATH`
<bav[m]>anadon: the build environment needs a python
<bav[m]>the "python" package doesn't supply this by default, iirc, but "python-wrapper" does
<anadon>bav[m]: That did fix the warnings! Now I just have errors I thought were tied to those warnings, but are apparently entirely unrelated X_X
<bav[m]>those patch-shebang warnings can be fairly catastrophic, because they often mean the script will not be able to execute
<anadon>bav[m]: did you say something? The freenode webclient kicked me.
<anadon>How do I make `/dev/urandom` available during tests? Or is it already? All I'm getting is that some filepath doesn't exist.
<raghavgururajan>bricewge, mbakke: Just letting you know that I have sent updated patch-set for #40994. I hope this set doesn't have any blockers for pushing. I have tried my best. :-)
<Kimapr_>how do you set up vpn server on guix?
***Kimapr_ is now known as Kimapr
<srandom>insmod wireguard.ko
<srandom>wg-quick up wgX
<rekado_>huh, I wonder why I was removed from the sudoers on bayfront
<rekado_>I was just going to upgrade the logging interface
<bdju>anadon: it doesn't look like you missed any messages, but there's a link in the topic to channel logs in case you want to double-check
*rekado_ updates
<Kimapr>is it possible to run openvpn server without certificates? all computers on my intranet are trusted
<Kimapr>without encryption*
<srandom>If you just feel that the certificate configuration of openvpn is very troublesome (it is really troublesome), you can use wireguard, the configuration and use are very simple
<Kimapr>but there seems to be no wireguard guix shepherd service
***apteryx is now known as Guest91892
***apteryx_ is now known as apteryx
<bricewge>Kimpar: Look here
<bricewge>I'm working on fixing wg-quick
<greenrd>Hi, I reported a bug via email to 3 weeks ago, but my report never showed up in the bug tracker or in the bug-guix mailing list archive
<greenrd>I notice that the mailing list archive is completely devoid of spam...
<greenrd>so seems like you have a really good anti-spam protection which I have to get past...
<reepca>greenrd: did you get an automated response? Subject would look like "bug #XXXXX: Acknowledgement: ..."
<greenrd>reepca: nope
<reepca>that's odd. It's been known to take a while to get propagated to the mailing lists for the first post, but the acknowledgement usually comes pretty quickly. I suppose you've probably already double-checked the "To" address?
<reepca>¯\_(ツ)_/¯ dunno then.
<greenrd>do I have to subscribe to the mailing list?
<bricewge>greenrd No need to subscribe to the ML
<greenrd>OK, but there is no spam on the ML, that I can see
<greenrd>so there must be some anti-spam system, if it's not "only subscribers can post"
<Kimapr>[#] ip link add wg0 type wireguard
<Kimapr>Error: Unknown device type.
<srandom>modprobe wireguard
<greenrd>maybe I need to set up SPF for my mailserver, so that it doesn't look spammy
<Kimapr>modprobe: FATAL: Module wireguard not found in directory /run/booted-system/kernel/lib/modules/5.4.35-gnu
<Kimapr>hmm, so i also have to install kernel module for wireguard
<srandom>...Okay, I ca n’t do anything about it. I did n’t use the kernel released by Guix, I used my own custom 5.6 kernel
<srandom>Guix seems to have a wireguard-linux-compat package, it may install the wireguard.ko module to extra/wireguard.ko, try to use an absolute path?
<Kimapr>i didn't install wireguard-linux-compat
<srandom>Try to install it, globally
<Kimapr>(kernel-loadable-modules (list wireguard-linux-compat))
<cbaines>hello o/
<sneek>cbaines, you have 1 message!
<sneek>cbaines, jsoo says: thanks for your patience with me. i'm working on purescript again.
<bricewge>Kimapr Yes reconfigure then reboot and modprobe will succeed
*sirmacik has put a patch proposal for cookbook inlusion but so far no answer >:
<bricewge>sirmacik: Thank you for that! I want to fix wg-quick before merging it
<sirmacik>oh, nice! (:
<sirmacik>than we'll be able to create wg servers with guix
<bricewge>sirmacik: I was slow at fixing wg-quick because I don't have a working wireguard server to play with
<bricewge>Could you try out a patchset that shoud fix wg-quick?
<sirmacik>yes, I can also give you acces to wireguard vpn later today
<bricewge>Thank you! That would be very usefull.
<bricewge>A lost a few days taming “guix deploy” to test this wireguard fix.
<sirmacik>bricewge: I'll try it after work
<sirmacik>now cannot disconnect from my vpn
<pinoaffe>I just realized that with the current autossh service definition, if several autossh instances are set up for a single user (which is fairly likely), they all write their pids to the same file - is this an issue, or are service PIDs read just once, sequentially, at startup?
<rekado_>pinoaffe: PID files can be used to check if a process is still running and it can be used to kill the process
<rekado_>if the files are overwritten that’s not going to work
<rekado_>the MNT Reform looks great
<rekado_>I’d love to build my own laptop around the board
<cbaines>I was looking at it, then I remembered I have a Pinebook pro hopefully arriving soon
<pinoaffe>rekado_: okay, so would you recommend altering the service definition so the pid-file location is user-configurable, or to automatically generate different pid file locations?
<bricewge>pinoaffe: It looks like a lot of services support specifying a pid file manually.
<pinoaffe>bricewge: I know, I just don't really like that solution all that much, since it offloads management to the user, and I don't really see a reason to expose pid files of services managed by shepherd to the user since the same info is accessible through the shepherd
<bricewge>I'm with you on that, such implementation details should be hidden away from users.
<bricewge>Such modification is far wider than just fixing autossh service though
<pinoaffe>for now I'll indeed just make the pid file configurable, but I don't think a more "hidden" intemplation need be very involved, a simple wrapper around `mktemp` (or a lambda doing something similar) should do the trick
<Noisytoot>What is the gnu/packages/python-xyz.scm file in the source code for?
<Noisytoot>I am not sure where to put a package
<rekado_>Noisytoot: python-xyz.scm used to be just python.scm, which contained *all* Python packages
<rekado_>we wanted to move Python itself to a separate module, so we kept it in python.scm and moved the rest to python-*.scm
<rekado_>most packages remain in python-xyz.scm, but it would be better to use the various python-*.scm if possible.
<pkill9>does anyone know of a python-pyqt application packaged in guix that uses QML?
<Noisytoot>rekado: Should a gopher client go in python-web.scm?
<bricewge>pkill9: Looks like carla use both python-pyqt and qtdeclarative
<bricewge>openshot and qutebrowser too . I found then using “guix refresh -l”
<pkill9>great, thanks
<mbakke>civodul: can I borrow your attention and insight for ? :-)
<sneek>Welcome back mbakke, you have 1 message!
<sneek>mbakke, lfam says: After a while on core-updates on my workstation, I realized that fonts are broken for Guix packages on my foreign distro (Debian) after the core-updates merge. <>
<mbakke>ooh, much nicer link
<civodul>Mumi is wonderful
<civodul>mbakke: why isn't the file writable in the first place?
<civodul>oh i see
<civodul>mbakke: isn't this root-owned-file-in-user-home situation a problem in general?
<civodul>i think it is
<civodul>independently of the patch stuff, if you do "sudo guix pull && guix pull", the second one likely fails with EACCES, no?
<jonsger>we dont have ldd in guix or?
<mbakke>jonsger: it's in glibc
<jonsger>mbakke: would it make sense to mention it in the description?
<efraim>I always grab it from gcc-toolchain
<mbakke>civodul: hmm, looks like you're right, the patch fixes build-self.scm, but the rest of the checkout is owned by root after using an inferior in 'sudo guix system reconfigure' :P
<mbakke>hm, but 'guix pull' still works, and afterwards no longer owned by uid 0
***jonsger1 is now known as jonsger
<civodul>mbakke: ah yes, "sudo guix pull" or "sudo guix system reconfigure" with an inferior will lead to a root-owned checkout
<civodul>but it's not new, right?
<civodul>or are you saying that somehow the problem doesn't occur with current master?
<PotentialUser-31>Hi, I want to install GuixSD with full-disk encryption on my librebooted x200, but I don’t want to enter my LUKS passphrase twice. On Parabola I put a key file inside the initramfs, but since GuixSD doesn’t use mkinitcpio or anything similar I don’t know how to solve my problem. Do you know of a secure (no key file in the store) workaround on GuixS
<civodul>PotentialUser-31: currently one has to enter the passphrase twice
<NieDzejkob>PotentialUser-31: It's not quite there yet, but not fundamentally impossible. A few patches would be needed...
<NieDzejkob>are you comfortable with Scheme?
<PotentialUser-31>unfortunately no.
*janneke is debuging set-network-interface-address
<mbakke>civodul: I'm not sure what's happening, but after a 'sudo -E guix system build config-with-inferior.scm', the files in the cached checkout are still owned by my user, except for build-self.scm and some obsolete/deleted files such as ham-radio.scm and a bunch of patches
<mbakke>perhaps libgit2 preserves user permissions? Can't find any evidence for it at a glance though.
<civodul>mbakke: ah yes, perhaps if the file are already there, it doesn't touch them, and thus ownership is preserved
<pkill9>why is it such a mission to get wrap-qt-program in a phase? it seems like you have to add a bunch fo modules to #:modules and #:imported-modules
<pkill9>wrap-program only requires an import at the top
<civodul>mbakke: but it's something you can't quite rely on, it depends on the previous state of the cached checkout, etc.
<mbakke>civodul: what's mind-blowing is that .git/index and .git/refs/heads/master also becomes owned by root; but after a new unprivileged 'guix system build config-with-inferior.scm', they are reset back to me as the owner
<mbakke>the only problematic file in all scenarios I've tested is build-self.scm
<pkill9>can anyone reproduce this error and tell me why it appears when building this package definition?: `guix build: error: file 'unquote-splicing/%python-build-system-modules.scm' could not be found in these directories: /gnu/store/pagj38w6fv83c82qfglk2vqbzjd637fk-guix-module-union/share/guile/site/3.0 /run/current-system/profile/share/guile/site/2.2 /gnu/store/5780x8w59lg898p9a45c2i18lx6r25yb-guile-next-3.0.2/share/
<pkill9>guile/3.0 /gnu/store/5780x8w59lg898p9a45c2i18lx6r25yb-guile-next-3.0.2/share/guile/site/3.0 /gnu/store/5780x8w59lg898p9a45c2i18lx6r25yb-guile-next-3.0.2/share/guile/site /gnu/store/5780x8w59lg898p9a45c2i18lx6r25yb-guile-next-3.0.2/share/guile /gnu/store/wrz50qkmv6cb58yzhxixyfcrpknp6vhr-pkill9-package-customizations/share/guile/site/3.0 /gnu/store/g9fjfkzk5mmyp3kjkd0mva7g3cslrw07-pkill9-nonfree/share/guile/site/3.0 /gnu/store/
<pkill9>sp38351jh5lwmzwr5m7a6rl5vx4cl5cs-pkill9-free-channel-dependency/share/guile/site/3.0 /gnu/store/bbnljryi501xzrl2nw0999qx4zwdm9qq-pkill9-free/share/guile/site/3.0`
<pkill9>i copied and pasted what the electron-cash package uses
<pkill9>the #:modules and #:import-moduel
<mbakke>pkill9: you need to start the arguments block with ` (quasiquote) instead of ' (quote) to allow for unquoting
<pkill9>ohhhh, thanks
<pkill9>i forgot i did that
<mbakke>civodul: also, just running 'guix pull' afterwards without privileges, with the root-owned patches, successfully resets the checkout and 'find ~/.cache/guix/checkouts/... -uid 0' returns nothing at all!
<mbakke>except for build-self.scm, if it has been modified
<mbakke>getfattr -d .git/index is empty
*mbakke questions his understanding of the Unix permissions model
<mbakke>derp, I need getfacl, not getfattr; though it did not solve the puzzle
*civodul had bricked Emacs due to an overly verbose REPL
<anadon>Can a package specify that something running as service, or is that reserved for system configuration?
<anadon> *runs
<mbakke>anadon: services must be declared in the configuration system, packages can not change anything outside of its /gnu/store/... prefix
<anadon>mbakke: How might I go about packaging a web service so that it is easy to start as a systemd service?
<mbakke>anadon: I don't know about systemd, but typically you'd package the service files and define a service that runs e.g. 'gunicorn /gnu/store/.../the-project.wsgi'
<mbakke>civodul: reading the trace, I see that libgit2 successfully does rename(".../index.lock" ".../index")
<mbakke>indeed, I can 'sudo touch test; touch test2; mv test2 test' just fine
<mbakke>renaming root files seems to be allowed when you own the directory
<mbakke>now, why does libgit2 fails to update build-self.scm, but not other files? stay tuned for the next episode (/me dinner time)
<anadon>mbakke: Is there an example I could look at?
<pkill9>guix should reject continuing the build if there's not even disk space
<pkill9>i mean, at the download stage
<Veera>raghavgururajan: Hi
<Veera>raghav-gururajan: Hi
<raghavgururajan>Veera: Hi! Congrats on Outreach for Linux Project. :-)
<Veera>raghavgururajan: Thanks! Congratulations for your selection to Guix Desktop Environment Project!
<Veera>raghavgururajan: I will hang on this channel for time!
<raghavgururajan>Veera: Thank you!
<raghavgururajan>That's great.
<Kimapr>weird. i changed commit of package definition for a kernel module and guix build didn't complain about checksum mismatch
<Kimapr>looks like a serious bug
<apteryx>are you sure you ran 'guix build' in your modified guix tree? (e.g., using ./pre-inst-env guix build) ?
<Veera>raghavgururajan: Did Linphone project made the break for selection!
<Kimapr>i don't have "./pre-inst-env" and have no idea where to get it
<Veera>Kimapr: Hi
<Veera>Kimapr: Do git clone of guix from savannah
<Veera>Kimapr: Then do ./pre-inst-env
<bricewge>Kimapr: Read at least that page and the next (14.1 and 14.2)
<raghavgururajan>Veera: I have no idea. ;-)
<Veera>raghavgururajan: Ah! Anyway great!
<Veera>raghavgururajan: It will be nice working and using Guix!
<pkill9>how many channels could be added to guix theoretically before it becomes slow to build?
<pkill9>ignoring the contents of those channels
<pkill9>so e.g. hypotheticlaly each channel had one simple package definition
<raghavgururajan>Veera: :-)
<ecbrown>pkill9: 1
*ecbrown ducks
<jonsger>mbakke: can you merge master again into core-updates?
<mbakke>alright, dinner finished, back to work
<mbakke>anadon: I don't have a specific example at hand, but many existing Guix services should be fairly similar (run this file with that executable)
<mbakke>jonsger: I'm thinking we can merge the other way around instead!
<mbakke>just sorting out a mind-blowing permissions issue first
<cbaines>mbakke, I'm not sure that permissions issue blocks merging core-updates
<cbaines>it's easily fixable if encountered
<mbakke>cbaines: I agree, I just want to finish the write-up first :P
<lfam>mbakke: I just replied to your questions about the fonts issue I'm having
<lfam>I'm still building to test the patching of fontconfig
<lfam>I did finally get a `guix pull` to finish with it, now doing the `guix upgrade`
<lfam>I'm going to try to just build hexchat and fontconfig for testing
<lfam>Actually inkscape has a smaller dependency graph, going to build that instead
<apteryx>looks like guix-set-emacs-environment (from emacs-guix) is broken
<apteryx>-> error: search-paths-specifications: unbound variable
<mbakke>lfam: there haven't been any updates to fontconfig this core-updates round, so I don't see what could be causing it
<lfam>mbakke: Yes, it's quite confusing. I strace-d inkscape and hexchat and didn't see anything obvious
<lfam>Do you have Guix on another distro? Can you reproduce it?
<lfam>I don't understand how applications find their fonts in general so I'm lost debugging
<lfam>Do they talk to fontconfig or something?
<slyfox>fc-list / fc-cache might be the query tools for current font cache
<mbakke>I don't have any graphical foreign system unfortunately.
<lfam>It's a bad regression if it happens on all foreign distros
<lfam>Can somebody running Guix on a foreign distro with graphical Guix applications such as Hexchat or Inkscape try `guix pull --branch=core-updates && guix upgrade` and see if those applications can still find the host system's fonts? You won't have to build much since there are substitutes available for core-updates
<mbakke>several users have reported that fonts were broken after switching to core-updates, and that it was fixed by installing font packages (they had none installed from before)
<mbakke>so I think something somewhere used to bundle fonts
<mbakke>pango perhaps?
<cbaines>is it obvious when running hexchat/inkscape that fonts are missing?
<lfam>cbaines: It's obvious for me in Hexchat because it looks totally different. And in inkscape, my drawings' text elements are missing
<lfam>Could be mbakke
<lfam>There are a couple things that don't make sense to me here. First, the regression, but if that can be worked around by installing Guix fonts... I suppose it's okay, although Guix apps should be able to use any fonts you ask them to IMO. Second, why don't the hard-coded gs-fonts in fontconfig work? And third, why doesn't our fontconfig successfully list fonts found in arbitrary directories?
<lfam>Like, I can use Guix's fc-cache to do `fc-cache ~/my-special-fonts`, and it says it finds them and caches them, but they don't appear in fc-list, or in Guix applications
<lfam>It could be that I'm mistaken and it never worked the way I thought, and something was bundling fonts
<lfam>It is fishy that our fontconfig doesn't successfully cache arbitrary fonts
<vagrantc>i definitely had two systems where i had never specified font-devjavu in user or system profiles, but installing it to the user profile fixed several programs
<vagrantc>that had previously worked before...
<lfam>vagrantc: Did that change occur after testing core-updates?
<lfam>And can you say what distro you are using?
<vagrantc>Guix System
<lfam>So it's not just on foreign distros
<vagrantc>so it's not just foreign distro
<lfam>(I don't have a graphical Guix System so I wouldn't have noticed there)
<lfam>Oof, my tests require building LLVM. It will be a while
<cbaines>I thought I had font issues after updating with core-updates, but the email I was reading was in bengali, I'd not paid the subject line enough attention
<cbaines>everything went fine for me in terms of fonts using Guix as a system, but I do have font-dejavu in my profile
<cbaines>I've got Guix on an Ubuntu laptop, and it's upgrading now, so I'll know how that works in a bit
<lfam>cbaines: Do your applications show any URW fonts available? Those are hardcoded in fontconfig and *should* work
<cbaines>lfam, on Guix or Ubuntu?
<lfam>Either cbaines
<lfam>I mean, they are hardcoded in Guix's fontconfig. I don't think Debian does that
<cbaines>If I open Gnome's fonts app on my Guix system, I see URW fonts among quite a few fonts
<lfam>I wish I had tested this sooner. I guess that Guix on other distros is not that common anymore
<mbakke>lfam: you can likely abuse grafts for your tests
<lfam>I'm feeling rushed as if this could be fixed in time
<mbakke>or (define fontconfig-for-test) and then replace the relevant inputs
<lfam>But it's a foundational dependency
<lfam>Even doing `guix pull`, I had to build many packages before I could even get output for --dry-run
<mbakke>lfam: adding --no-grafts should fix that
<lfam>I'd also hoped to fix the alsa-lib problem this cycle but oh well
<lfam>I'm finding it hard to focus lately
<mbakke>lfam: I doubt we'll figure out the font issue today, so I'm considering doing the merge for more visibility and testing
<lfam>Yeah, that's the right thing to do
<mbakke>lfam: alsa-lib can probably be fixed on 'staging' at least
<mbakke>cool :-)
<apteryx>mbakke: FWIW, I've been using core-updates for about a week, other that what I've already reported.
<apteryx>*without any issues
<mbakke>apteryx: good to know :-)
*mbakke starts the merge
*jonsger will pull the new master as soon as possible :)
<lfam>So that patch on fontconfig didn't help
<mbakke>lfam: can you try downgrading pango?
<mbakke>'make check' passed!
<mbakke>951 commits, wow
<mbakke>jonsger: grab it while it's hot! :-)
<lfam>That's a lot of new LOC
<apteryx>mbakke: thank you for leading the effort!
<apteryx>by the way, I've recently started seing warnings such as: In procedure load-thunk-from-memory: incompatible bytecode kind when building the Guix tree... Am I doing something wrong?
<apteryx>Would a 'make clean-go' resolve the issue?
<mbakke>apteryx: that means you have .go files built with a different version of Guile
<mbakke>apteryx: you are welcome, I hope more people will contribute to the branch in the future though ;-)
<apteryx>ack :-)
<apteryx>another question: is building the tree with Guile 3 now something that is expected to work without glitches?
<mbakke>I think that has been supported since 'guix pull' was changed to use Guile 3
<apteryx>OK, I'll give it a shot
<apteryx>and compare compiling times
<lfam>apteryx: That would be really valuable info
<jonsger>mbakke: thanks!
<jonsger>guix-translated-texinfo takes quite some time. Is this normal?
<apteryx>jonsger: it does!
<apteryx>ah, it seems I must have been building Guix with Guile 3 for some time already, given that 'guix environment guix' pulls guile@3.0.2
<raghavgururajan>apteryx: I just saw mbakke's commit 4bdf4182fe080c3409f6ef9b410146b67cfa2595. Does this mean, linphoneqt is in master now?
<apteryx>raghavgururajan: yes!
<raghavgururajan>apteryx: Yay! Thanks mbakke.
*raghavgururajan guix pulls
*rekado_ still works on the lightdm service review
<mbakke>raghavgururajan: you're welcome, and thanks for all the new packages lately! :-)
<civodul>jonsger: it's sequential + po4a is extremely slow
<raghavgururajan>mbakke: Cheers!
<raghavgururajan>Ooo, building linux-libre.
*lfam goes back to trying to fix alsa-lib for foreign distros
<wtx>hi all
*raghavgururajan does Ctrl^c
<wtx>is normal that guix takes more than five hours for install icecat?
<wtx>is still working
<wtx>the machine is some old, its a netbook
<cbaines>wtx, installing icecat should be quick, you're probably building it as well
<raghavgururajan>wtx: Are you on Guix System or Foriegn Distro?
<cbaines>which could easily take hours
<raghavgururajan>wtx: Ah, please make sure substitutes are enabled.
<cbaines>raghavgururajan, I'm not sure that will necessarily help at the moment
<wtx>at this moment is building rust-1.19.0
<cbaines>Looks like there isn't a substitute available for the two most recent icecat outputs from
<wtx>i use "package -i icecat" i think
<raghavgururajan>cbaines: Substitutes are disables by default on foreign distros correct?
<cbaines>raghavgururajan, I don't believe so, if you use the install script, it asks you if you want to enable them
<wtx>architecture is i386
<raghavgururajan>I see. I doubted about susbtitutes being enables because, I just installed icecat.
<raghavgururajan>Ah, different Arch. My bad.
<pkill9>yea i wouldn't bother building icecat, especially if you're also building rust
<pkill9>i would just wait for a substitute
<wtx>it's crazy
<wtx>so what i have to do?
<wtx>should i have to break the building?
<cbaines>If you want to install Icecat, I'd recommend letting it complete
<pkill9>that will take forever on a netbook though
<lfam>Especially an i386 netbook
<wtx>but cpu is running now about 10 percent
<cbaines>Assuming i686 is the system mapping to i386, I'm not sure there's been a substitute for icecat for a while
<cbaines>at least the Cuirass builds seem to have failed for a while
<wtx>you are right
<wtx>i am running under i686
<lfam>I doubt it will complete successfully
<wtx>maybe tomorrow
<wtx>or next week
<mroh>hurray! merge happened!
<drakonis>an interesting tidbit
<drakonis>the founder of hashicorp is now running nix
<mbakke>wtx: rust does not build on i686 currently, sorry for the wasted CPU cycles :-/
<wtx>thats why the cpu is in 10 percent usage
<wtx>it gets all versions of rust
<wtx>from the newest to oldest
<rekado_>I only know hashicorp from one of civodul’s talks; a quote about adding an exciting new feature to a DSL: loops.
<lfam>They made some neat free software tools for deployment / orchestration
<lfam>mbakke: Should we restrict the set of supported-systems for Rust?
<mbakke>lfam: yes, probably
<wtx>it's impossible then to run icecat on this machine ?
<rekado_>ah, I suppose I’m just not in their target audience
<lfam>wtx: It's not possible to use Icecat from Guix on your computer. We don't support your CPU type for packages that use Rust
<lfam>Maybe in the future we will add support. It depends on if anyone does the work to make it happen.
<wtx>i know
<wtx>i386 is devaluated
<lfam>It's true
<rekado_>I still have a i686 machine that I use daily
<rekado_>just not with a browser
<cbaines>Well, i686 is a supported system for Guix, it's just that there's some issues with rust and that causes problems with icecat
<wtx>thats why is break in building rust
<wtx>there is a warning i suposse that it havent relation
<rekado_>hmm, now that has all the Debbugs data and a local db I think I can add new issue selectors, such as “forgotten” issues (open issues with last message long in the past) and “lonely” issues (issues where only the submitter has sent messages).
<wtx>LC_ALL cannot change locale
<wtx>no relation with that
<wtx>Ctrl^C then
<rekado_>I hope that by adding new selectors we can better rediscover forgotten or accidentally ignored submissions
<wtx>any recomendation then for a browser
<wtx>for this machine
<rekado_>wtx: eolie and epiphany are Webkit-based browsers
<jonsger>rekado_: what do you think about a mailto: link near to the comment section. So you could use your own mail client/smtp server?
<vagrantc>netsurf is pretty lightweight ... but a little rough around the edges
<vagrantc>wtx: ^^
<wtx>i know netsurf
<wtx>its very lightweight
<wtx>also i know epiphany
<rekado_>jonsger: when the commenting feature is enabled the mailto: link is hidden. I also think it would be good to display it anyway.
<wtx>not eolie
<rekado_>there’s also “next”
<rekado_>and “nomad”, but I don’t know if there’s a package for it yet
<jonsger>wtx: chromium seems to built for i686
*rekado_ keeps forgetting that chromium exists
*cbaines doesn't when it begins to build
<wtx>yeah but chromium is g* shit
***michel_slm[m] is now known as michel_slmOld[m]
<cbaines>I have to remember to build with less than all cores, because at points it requires quite a lot of ram per core
<rekado_>yeah, it is google’s hit browser
<jonsger>but Icecat has problems with video conferencing
<lfam>wtx: Our package of Chromium removes the problematic parts
<rekado_>…or does video conferencing have problems with icecat?
<lfam>It's called 'ungoogled-chromium'
<apteryx>rekado_: it works mostly fine for me, I think the only thing Chromium for WebRTC that Icecat (firefox) doesn't is RTX, something that sends the audio (and video?) streams in a way that guards against breaking audio on congested networks.
<wtx>are there precompiled chromium in guix?
<jonsger>wtx: yes
<wtx>package -i chromium
<jonsger>wtx: ungoogled-chromium
<jonsger>apteryx: the problem is that Icecat is ESR so we are missing some patches for video conferencing from upstream, I guess
<apteryx>that said, I find Icecat to work much more efficiently on my system. For some reason Chromium seems to like using the disk a lot, even when there are still 2 GiB of RAM free.
<apteryx>there is*
<wtx>maybe the disk work are g* bd
<rekado_>I only use ungoogled-chromium once a week (and only since the pandemic) for work meetings
<rekado_>I’m okay with icecat, even though it has a few annoying quirks
<rekado_>(like the way HTML is formatted and edited in the inspector)
<jonsger>and icecat has problems playing h264 videos. Chromium is fine
<apteryx>also, icecat's search function is orders faster than Chromium's for some reason. Try it on a multi-megabyte text page.
<apteryx>(it's striking for example when attempting to search through a build output on Jenkins)
<wtx>i dont like chromium
<wtx>even i think firefox-esr is faster than chromium
<wtx>and the videos and streaming are not a proble
<wtx>at least for me
<wtx>the building is freeze
<wtx>i am knoking with Ctrl-C
<wtx>no answer :|
<jackhill>civodul: Did you see my reply about emacs-next? (I replied to a closed issue, so I don't want it get lost). tl;dr with your change to the patch the addition of the (ice-9 match) import is not longer needed.
<lfam>jackhill: civodul is probably gone for the night. You can use the sneek bot to relay the message when he is back
<lfam>sneek: later tell jackhill: You can use sneek like this
<sneek>Got it.
***michel_slmOld[m] is now known as michel_slm
<jackhill>lfam: thanks
<sneek>Welcome back jackhill, you have 1 message!
<sneek>jackhill, lfam says: You can use sneek like this
<jackhill>sneek: botsnack
<lfam>You already knew :)
<jackhill>sneek: later tell civodul: Did you see my reply about emacs-next? (I replied to a closed issue, so I don't want it get lost). tl;dr with your change to the patch the addition of the (ice-9 match) import is not longer needed.
<sneek>Will do.
<jackhill>lfam: yes, but I forgot about timezones :)