IRC channel logs
2023-05-03.log
back to list of logs
<jonsger>ACTION still tries to update telegram-desktop to 2.8.1... <sneek>tschilptschilp23, you have 1 message! <sneek>tschilptschilp23, apteryx says: audacity should be fixed; 3.3.0-beta-1 <bjc>is anyone here using pipewire and wouldn't mind sharing the important bits of their configuration? <old>apteryx: could it be because my home profile is not up to date? <old>I had failed builds that I fixed lately that was preventing from updating home <atka>how minimal can guix system be, kernel, shepherd, bash, coreutils small? <atka>plus guix and required tooling I suppose <vagrantc>minimal is not something i associate with guix :) <atka>hi vagrantc, yeah I know <atka>I'm looking/hoping for an alpine replacement that is immutable/declarative <atka>the nix folks say the same, nix is bloated <atka>vagrantc: how's the solar experimenting going? <mirai>what kind of “bloat” are you trying to avoid? <atka>minimal packages, size, low ram requirements <atka>basically, kernel, systemd/shepherd, busybox or coreutils <atka>usable on systems with 1gb ram <atka>alpine and void fit the bill perfectly, but lack the declarative nature and immutability <vagrantc>atka: solar goes mostly well, though i have yet to integrate guix into it in any way :) <atka>if guix could run on a pi and be small enough I'd have guix + grafana going <atka>I'm at 11 months with my system in alaska <atka>but could be a lot more, just don't have anywhere for it to go <atka>700W of poly panels, victron mppt, 1280Wh battery storage <atka>powers everything completely off grid, December and early January is rough though <atka>vagrantc: how is life with the honeycomb? <zamfofex>Hello, Guix! Sorry for nagging, but would anyone mind taking a look at my patch <https://issues.guix.gnu.org/63088> (adding Lc0, a chess engine)? I don’t know if it’s inpolite to ask like this. I was trying to avoid asking while there were unreviewed additions before mine, but now that it’s off the <https://issues.guix.gnu.org/recent> page, I can’t really easily verify that anymore. <podiki[m]>btw guixers, I have godot 4 working locally, still want to clean up/debundle a little more if I can before submitting <atka>nice work podiki[m] godot would be nice to have <podiki[m]>we have 3.3 I think? so probably should update that to I think 3.5 and keep it as the lts release <oriansj>apteryx: the prompt was "tell me the story of how guix came from a 357byte bootstrap seed" <oriansj>the funnest part was it then tried to convince me that it was a closely guarded secret locked away and guarded with lethal protections. And here I am just thinking "Mother F&^ker it is on a half dozen public websites" <oriansj>anyone know how to fix: guix shell: error: statfs: : No such file or directory <oriansj>apteryx: it was hit when running: guix shell -CFN --preserve='^DISPLAY$' --preserve='^XAUTHORITY$' --expose=$XAUTHORITY --preserve='XDG_RUNTIME_DIR' --expose=/run/user/"$(id -u)"/pulse alsa-lib bash coreutils dbus-glib file grep gtk+ libcxx libgccjit pciutils pulseaudio sed -- ./tor-browser/Browser/start-tor-browser -v <apteryx>perhaps you need to expose the whole of XDG_RUNTIME_DIR? <podiki[m]>maybe something funky with one of the env variables there? <FloraGordon[m]>!!! Enjoy the most profitable financial market (crypto market ) as you get 100% profit...and you can also make up to $100k or more in 3days send me a private message and ask me HOW on TG <cedb>i used the guix install script but cant find any of the source code anywhere in my system for the packages, just the store stuff, i might have used "authorize substitutes" when installing but i cant see how to "deactivate it" <cedb>ive tried 'sudo guix pull --no-substitutes', doesnt seem to chagne anything <abrenon>first attempt to update my system since the big core-updates merge, I notice that there's no substitutes for csvkit and indeed it fails to build locally <cbaines>bost, have you tried the suggestion to use the rest backend? <bost>cbaines: uh, sorry. I stopped thinking after seeing "This is [..] a bug in the server". Ok. I'm going to try the suggestion. <ngz>Hello. I'm unable to push commits on the repository. I get "no code for module (git)" and "make: *** [Makefile:6943 : authenticate] Erreur 1". What am I missing? <bost>cbaines: yeah, but there's bit of work for you, too :) <bost>cbaines: Ok. I'm gonna mention it in the github issue. <reyman>i have a manifest that load an inferior to access an old version of pytest package. <reyman>this is needed by python-markdownify that say pytest<7,>=6.2 <reyman>but even with the inferior specifying the good package, markdownify build say : ERROR: Could not find a version that satisfies the requirement pytest<7,>=6.2 (from versions: none) <next4th>reyman: python packages are specified to the python version it built with (installed into lib/python-x.y/site-packages), that maybe the reason. <cbaines>reyman, are you expecting python-markdownify to be built with python-pytest@6.2.5 ? <cbaines>because your manifest doesn't specify that <cbaines>you need to replace the python-pytest package in the inputs/native-inputs of the python-markdownify package, rather than including it in the manifest <cbaines>reyman, this isn't something you should be trying to workaround though. The package definition for python-markdownify in Guix is broken, and should be fixed <reyman>so i need to inherit and override (redefine) the actual definition of python-markdownify? <cbaines>yep, but I would try and avoid doing this if possible <reyman>ok. so the best at this point is using time machine to build my package before the upgrade of python pytest <cbaines>reyman, it's the recent merge of core-updates that introduced this problem, so if you can use this commit of Guix 472706ae2f9160833951a4e4bcc4c206e03097b0 (or an earlier one), you shouldn't have this problem <next4th>look like update it to 0.11.6 is okay, i'm doing it <PotentialUser-85>Hello. I wonder if there is an option to check a complete channel for CVEs with guix lint instead just a single package <reyman>i have also a broken package of python yubikey with latest pull <next4th>reyman: python-markdownify fixed in master now, thank you for report :) <sepi>I'm having trouble guix pulling with a channel I created recently. I get the error https://paste.debian.net/1279153 . The error in the derivation log is "no code for module (mnhnl packages python-expocms)". <sepi>I could use that code before I transformed it into a channel by using -L and guix system reconfigure <sepi>Any idea what could be the problem here? <ngz>Hmmm. I'm still experiencing "compilation of /home/ngz/dev/guix/./build-aux/check-channel-news.scm failed" and "no code for module (git)" whenever I try to push commits, even after a make clean. <sepi>Nevermind, I had forgotten to push code to my channel where the module hierarchy was broken due to renaming. <cbaines>ngz, does (use-modules (git)) work in a Guile repl? <ngz>cbains: In guix repl, I can (use-modules (git)) without problem. <cbaines>ngz, ok, but what about my question? <cbaines>(since I don't believe check-channel-news runs in a guix repl) <ngz>cbaines: Sorry, I misread your question. The error is the same in a Guile REPL. <ngz>(no code for module git, that is) <cbaines>ngz, no problem, so I use "guix environment guix" when hacking on guix to provide the development dependencies, and this does provide guile-git <cbaines>I'm guessing you don't, so what approach do you use for the things you need for hacking on guix? <oriansj>apteryx: finally figured it out. XAUTHORITY doesn't exist under sway <oriansj>guix probably could benefit from a better error message when --preserve= is given nothing rather then just "guix shell: error: statfs: : No such file or directory" <ngz>cbaines: the problem is not hacking on Guix, where I use "guix shell -D guix". Here I try to push local changes upstream. <ngz>cbaines: I'm suspecting the pre-push hook <cbaines>ngz, it will be the pre-push hook, could you do the push inside guix shell -D guix? <cbaines>you could also try tweaking the pre-push hook to use guix shell <ngz>cbaines: OK. This works. But I cannot switch to "guix shell -D guix" environment each time I want to push something because I use Magit from Emacs. <cbaines>ngz, I use magit too, but I use direnv and emacs-direnv so that the environment I get in a shell is the same in Emacs <ngz>cbaines: I never used these. <cbaines>it's not a perfect approach, emacs will hang while guix takes it's time doing stuff, but it's been working for me <cbaines>as I say, an alternative is to alter the pre-push hook so that it does the guix shell bit <ngz>I used to tweak pre-push hook, but I think recent changes to that part of the code removed them. <ngz>I need to add "guix shell -D guix" just before "make authenticate …", right? <ngz>"exec guix shell -D guix" I mean. <ngz>cbaines: OK. Thank you for your help! <cbaines>ngz, the other approach might be to do something like eval `guix shell -D guix --search-paths` <cbaines>if you do that before running exec make, then I think that might work too <cbaines>it might be a bit neater that way since then there's not an extra process running <ngz>OK, I'll try for my next push <apteryx>oriansj: oh, nice find. could you please report the subpar error reporting bug? <apteryx>anyone has opinion to switch the default cups package in cups-configuration from cups to cups-minimal? <apteryx>lechner: I'm still wondering about friendly-bot; where did they go? does it crash every couple days? it was useful :-) <guikas>hello people, i need some help on correctly configuring multiple keyboard. i read the the manual `https://guix.gnu.org/manual/en/guix.html#Keyboard-Layout`, but there is no mention about how to set _multiple_ keyboard layouts.i'm assuming simply doing something like this will work (i also map caps to ctrl):(keyboard-layout "us" <guikas> (xkb-options '("caps:ctrl_modifier"))) <guikas> (xkb-options '("caps:ctrl_modifier"))) <sepi>When executing plain guile on guix system, is it supposed to be able to load packages from channels without specifying -L? <apteryx>if you launch it via 'guix repl' yes <apteryx>if you just (use-modules (gnu packages something)) I'm not sure <sepi>aaa, I'll give that a try <sepi>apteryx: nice, that works. Thanks! <sepi>Is there a place that documents the whole indirections (symlinks) and directory structure of profiles? <teddd>Is it necessary to run make before using ./pre-inst-env ? make takes ages on my machine <rekado>if you don’t run make you’ll waste even more time evaluating everything again and again <teddd>Is it possible to just recompile the files I changed then ? <teddd>what exactly do you mean by evaluating again and again ? <Parnikkapore_m>if you don't run make, it will interpret the Guix code instead of running the bytecode version, which is even slower <Parnikkapore_m>I don't know a way use the .go-s from the in-Store Guix inside pre-inst-env unfortunately <teddd>I see. So the workflow to change something in the source tree is: change things, run make, wait, run pre-inst-env, check results, repeat <Parnikkapore_m>you could, but generally after git-pulling I run make once and just go to town <Parnikkapore_m>in that case only the files you changed get interpreted, which is fast enough <teddd>I see. That sounds good. I'll try that. Thanks :) <stevenroose>Why is there no substitute for the rust package? Weather says "0.0% (0 out of 2) of the missing items are queued". It's been like this for several days, I've been waiting on it. <stevenroose>Is CI just a lot behind? There used to be 1000 queued items, which I thought was maybe the limit, but not there's only 889 and still rust is not queued. <cbaines>stevenroose, you should be able to get a substitute for x86_64-linux at least <ngz>What is the process to handle patches implying world-rebuild nowadays (e.g. bug#63124)? <cbaines>can you share the specific derivation (e.g. guix build -d --no-grafts rust) <ngz>I'm fine with the idea. Would the branch be associated to a qa process? <civodul>i'm not sure about qa.guix; we'd need help from cbaines <cbaines>all branches get processed, but currently submitting builds for branches is a manual tax <bjc>if we're going to start setting up branch builds, it'd be nice if we could integrate debbugs into them; so that a given patch isn't misapplied to master before automated building <cbaines>bjc, what do you mean by misapplied? <bjc>if i have a patch for rust-updates, it shouldn't get applied to master <cbaines>in this hypothetical case, what means the patch should be applied to some branch and not master <civodul>ngz: so i guess you can do that: create the branch, apply those two patch series, perhaps let guix-devel know in case some other texlive patches are pending, and then ping us so we get it built <bjc>an open question, but maybe just a regexp on the subject: ‘\s+([^]]+)\\]’ or something <cbaines>bjc, I'm not sure we want to force all changes in to non-master branches <bjc>so that if i send a message ‘[PATCH foobar]’, and we have a branch ‘foobar’, that'll get used as a base <bjc>it would only be for explicitly tagged patches. default should still be master <cbaines>I think there's an assumption here that these branches aren't going to overlap much, but I'm doubtful (or at least unsure how to check) <stevenroose>cbaines: /gnu/store/daw7ddsinyh10y6q695hlmxnxfah38y6-rust-1.65.0.drv <stevenroose>cbaines: I just do `guix weather rust` and it says (for x86_64-linux) 0% of packages available. From both ci. and bordeaux. <cbaines>stevenroose, what revision of Guix are you on? (guix describe will show you)? <cbaines>The data service doesn't know about that derivation <efraim>I'm about to refresh the branch for rust-team <bjc>i'm not really concerned with overlap. i'm concerned that if i'm working on a relatively long lived branch, like we're discussing here for texlive, then it's a waste of time to apply a patch specifically for that branch to master <stevenroose>cbaines: Generation 26 Apr 30 2023 20:56:39 (current) -- guix e132e48 <efraim>I got rust-bootstrap to build on aarch64-linux <ngz>civodul: this sounds like a plan. I'll start by applying 63143 into a new branch, e.g., tex-team, and push it. <stevenroose>Did a guix pull right before I started waiting on that rust package. Didn't pull in more recently because I thought I wasn't getting the substitute because the package was too new. Didn't want even newer. <stevenroose>I also didn't get the linux substitute from nonguix but after 2 days waiting it was available eventually. <cbaines>stevenroose, I only see rust@1.60.0 for that revision (e132e48) <efraim>stevenroose: it's a branch on the savannah repo <stevenroose>cbaines: I just did guix package -u rust and it went for 1.65.. nonguix doesn't add another branch of rust versions right? <stevenroose>cbaines: guix package -s rust shows both 1.60 and 1.65 for me <civodul>the other texlive patches are much needed: those deprecation warnings when running "guix pull" suck <cbaines>stevenroose, hmm, I'm unsure, I do see rust-1.65, but it's not exported from (gnu packages rust) <stevenroose>cbaines: try guix package -u rust, will it not start building v1.65? <stevenroose>My current rust is broken, rustc can't build anything because the link with glibc is broken.. <cbaines>stevenroose, nope, I get rust@1.60.0 <stevenroose>It wants glibc 1.34 but a recent Guix system update only leaves me with glibc 1.35 <efraim>looks like there's problems with the git hook and git worktrees <stevenroose>cbaines: hmmm, maybe I once did guix package -i rust@1.65 or something and now it registered that I want that version? <stevenroose>But so the 1.65 package is not available as a substitute. Is that expected behavior then? <stevenroose>Without a substitute, it builds 1.61, 1.62, .., 1.65, which takes ages.. <guikas`>just curious why you guys dont want to use something slack or discord or even matrix (as i remember it's open source)? keeping my irc client open just for keeping track of chat sounds pretty cumbersome ngl. <cbaines>stevenroose, I'm not sure either of the default substitute servers will provide substitutes for these rust packages <cbaines>the definitions are in Guix, but they're not exported in the (gnu packages rust) module, so they won't be picked up as things to build <stevenroose>cbaines: is there a way to see if I somehow manually forced guix to want v1.65? <cbaines>stevenroose, there might be, but I'm unsure <cbaines>efraim, do you know why there is a rust-1.65 package definition? <efraim>people wanted to get a start on packaging the next version of rust <efraim>ok, new version of the rust-team branch is now pushed <cbaines>efraim, but why not make it public if the package definition is there? <efraim>it's built without any of the tests, and IIRC we wanted to only have 1 supported version of rust at a time <efraim>plus it can still be refered to with @@ if needed in other channels <stevenroose>Is there a doc on how the rust-team branch works and how I use it? <efraim>there isn't, it's like the other feature branches, but it's taking a bit longer since the core-updates merge broke rust on aarch64 and riscv64 <stevenroose>I don't know how feature branches work, but I guess it's a dev branch, not supposed to be used as a channel to get different rust package versions, right? <efraim>yeah, basically. it's so changes which cause many rebuilds can be staged and then brought into master after it's been built out by CI <stevenroose>If anyone has any suggestions as to how to fix my current rust that is broken because of glibc, please advise :) <guikas`>sorry for interuping again, but how to check if there is a version of birary package. i want to know if there is mainline binrary version of linux kernel, cause i dont want to compile linux from source, cause my pc is potato <stevenroose>I get this: /var/cargo-target/debug/build/slab-e701ba5622adaa00/build-script-build: /gnu/store/5h2w4qi9hk1qzzgi1w83220ydslinr4s-glibc-2.33/lib/libc.so.6: version `GLIBC_2.34' not found (required by /var/cargo-target/debug/build/slab-e701ba5622adaa00/build-script-build) <cbaines>guikas`, that's offtopic for this channel since guix doesn't have a package for linux, and uses linux-libre instead <stevenroose>And I saw in the repo that 2.34 was replaced with 2.35. <bjc>for linux-libre: guix weather linux-libre <stevenroose>bjc: why does guix weather by default not try all configured substitute servers? <bjc>it does. typically we have two: ci and bordeaux <stevenroose>I have 2 others configured, nonguix and nonguix-gaming and it doesn't check though, only when I use manual --substitute-urls= argument. <cbaines>guix weather doesn't ask the guix-dameon to check the substitutes, instead it checks directly itself <bjc>oh, i'm mistaken then. i only use the two, and i get results for them <cbaines>so it doesn't use the guix-daemon configuration for substitutes <stevenroose>Oh, is there another way to configure substitutes? Yeah I use the guix-service-config way in desktop-services. <cbaines>stevenroose, while I can't claim to know exactly what guix weather should do, one key use is to check the substitute availability from the default substitute servers, which it does just fine <cbaines>I'm not sure if it's meant to be used by people to check the substitute availability of individual packages that they're interested in <guikas`>computing 23,337 package derivations for x86_64-linux... <stevenroose>cbaines: "Report the availability of substitutes." is what the help page says :) <stevenroose>Might be opinionated but at least this is what I consider expected behavior. <ngz>civodul, rekado: I think the origin/tex-team branch is now available. <ngz>Ewww there are around 80 texlive packages to rename. <apteryx>guikas`: IRC has pretty much the lowest technical requirements to get started, works reliably with little resources. I can't even create a Discord account because it requires a mobile phone number (I use VoIP). <barbaneigro>hey, does anyone has some tips on how to get guix-daemon to autostar in a selinux distro as fedora. tx <acrow>stevenroose: Reading through your posts it might be worth a try for you to do a guix system reconfigure following your pull of the new core-updates. The substitute you want may require the new glibc but your system is still calling for the prior version. After the core-updates I had some funny things happen and they all corrected themselves after I did a system reconfiguration. Hopefully that will take less time than rebuilding rust, <acrow>stevenroose: and the reboot that follows. :) <podiki[m]>on branches: am I able to create on one savannah or can only some committers do that? and for having it built I guess needs a ping to the cuirass wranglers <ngz>podiki: You need commit rights to create a branch on the repository. Then, you need to ping developers here to create a jobset on the CI, indeed. <podiki[m]>Right so any commiter can create, no force pushing though (or is that just for master)? <ngz>At some point, the branch will probably be merged into master, so the history should be clean <ngz>civodul: Thanks, too! However, I'm planning to do some work there in the following days, so it may not be useful to burn CPU cycles on this branch yet. WDYT? <civodul>i clicked on "Deactivate" but i'm not sure what will happen <podiki[m]>ngz: makes sense, thanks; I'll make sure before creating <sepi>How can I get the shepherd to log its exceptions when failing to start a service? I don't see anything on /var/log/messages <sepi>When running # herd start gunicorn, I get herd: exception caught while executing 'start' on service 'gunicorn': Wrong type to apply: 81 <efraim>sepi: sounds like the start command is wrong <sepi>efraim: yeah, you might be right but how can I debug it :/ <sepi>attila_lendvai: how can I get your branch into my guix system? <unmatched-paren>i apologise for the lack of work on dissecting guix part 4; i've been taking a break from guix stuff for a bit <old>Is there a way to subscribe to certain packages on cuirass? <old>I would like to subscribe to every packages in gnu/packages/instrumentation.scm so I can fix them whenever there's a break <sepi>efraim: also, do you understand that "Wrong type to apply: xxx" message? I get it in different variations but don't really get it <bjc>it means you have something that isn't a function in the function position of a form, like: (18 foo bar) <bjc>how'd it get there? i dunno. probably missing a quote or the use of ‘list’ <sepi>bjc: ok, then my interpretation is right. It seems though that it's not directly me producing this form <sepi>also it shows either numbers (maybe chars) or whole strings as being the object that is being evaluated <bjc>i'd guess there's some macro expansion causing problems. it can be annoying to track down, unfortunately <bjc>one of the many reasons i dislike macros <sepi>bjc: it could be the PID of the process that exits prematurely :) <sepi>bjc: I'm slowly getting there :) The at least problems with the processes being called badly can be debugged using syslog <elevenkb>hey y'all's I would like to use the crocus driver <kaelyn>elevenkb: the crocus driver isn't supported quite yet as it requires updating the mesa package definition (which will first require a feature branch for testing and substitute building). https://issues.guix.gnu.org/63219 contains a set of patches for the forthcoming mesa feature branch, including enabling the crocus driver. <elevenkb>kaelyn: thanks for letting me know! the issue is that sway doesn't launch properly for me now. I guess that I'll have to try fixing the version of sway then. <bjc>ACTION peers at sneek <kaelyn>elevenkb: I suspect your best bet may be to "guix pull" to a commit shortly before the core-updates merge (or roll back to a recent generation from before the merge) so that the older mesa driver will be available for your video card. <mirai>apteryx: is there a way to easily fetch the patch series ? <apteryx>then from the summary buffer, I input: C-u 9 | <apteryx>and choose 'git am -3' as the command <apteryx>the current directory of your main Debbugs buffer (e.g., *Guix Patches*) should be set to your Guix checkout, with M-x cd <Altadil>Hi, in the Getting Started section, the manual mentions the need to run "hash guix" after a guix pull. I’m unclear as to whether this applies to Guix System, or only to guix on foreign distros ? <zacchae[m]>How should one deal with packageing python modules that require a non-python build-system like cmake? I'm thinking I would create an intermediate package building with cmake, then try to build the python package from that? <ade>hey guys have anyone encountered this error when reconfiguring guix system?: <ade>guix system: error: symlink: Permission denied: "/var/guix/profiles/system-4-link.new" <ade>in documentation i rember using guix should not require sudo <Guest19>"docker-compose up" returns "/gnu/store/75sk24gki8b4w9lblyhdn5m4dw3ma55r-python-paramiko-2.7.2/lib/python3.9/site-packages/paramiko/transport.py:219: CryptographyDeprecationWarning: Blowfish has been deprecated <Guest19>though my Guix is roughly 2 weeks old. Just wanted it to mention <mekeor[m]>Guest19: i doubt that just mentioning an issue here, in the irc channel, ensures that the community acknowledges your issue. i guess, sending an email to one of the mailing-lists would rather do so <bdju>did my email to bug-guix about dolphin-emu failing to build not go through? I sent it but never got the confirmation email with a bug # <mekeor[m]>bdju: if you're not member of the mailing-list, afaik, an admin needs to manually approve it, which might take a while <bdju>I have sent many reports before <bdju>and I am subscribed to the mailing list as well <mekeor[m]>does debbugs normally send acknowledgement-mails? <bdju>2021-05-11 07:12 AM GNU bug Tracking ! bug#48347: Acknowledgement (missing debug symbols for qutebrowser and python packages) <bdju>also I didn't see my own bug appear via the mailing list which also should've happened. so that's 2 emails fewer than normal <bdju>(I've sent more recent bugs in than that but my mail client may not be the best at searching old things) <nikita>Requires: A list of packages required by this package. The versions of these packages may be specified using the comparison operators =, <, >, <= or >=. <nikita>my guess would have been "=" instead of "==" without looking at that <podiki[m]>elevenkb: re: crocus driver, yes saw that reported (maybe you?) and is in the mesa series of patches i will review and hopefully have a branch built in the near future <podiki[m]>how does one specify libstdc++ from the cli? e.g. guix build -e "(@ (gnu packages gcc) (make-libstdc++ gcc)" errors with failed to evaluate expression <podiki[m]>ah, can do with "(@@ (gnu packages gcc) libstdc++)" for the default <gabber>i have troubles testing my generalization of a home-night-time-service type (https://termbin.com/flybg). i build test images but the service type does not seem to exist in the built system. <gabber>i enter a `guix repl` and (use-module (gnu home services desktop)) but i get "error: home-night-time-service-type: unbound variable" <gabber>needless to say i'm unable to `guix home reconfigure` (with the same error) <jacereda>Hi, anyone else having problems with gdm/wayland recently? It just shows the cursor on a black screen for me... <jacereda>My config works fine via `guix system vm` <gabber>i couldn't say that i have.. mine works just fine. the "cursor and black screen" is when GDM asks for your login credentials? <gabber>does it change when you hit a key? and if it is a laptop: does it help if you suspend it (close the lid) and wake it up again? <Guest61>jacereda: can you switch to another tty? I recently had this on my computer after a kernel switch <Guest61>Does anyone have an idea how it can happen that erc times out with '==> ERROR from irc.libera.chat: Closing Link: 127.0.0.1 (Connection timed out)' ? <Guest61>I double-checked my config, and honestly haven't changed anything as far as I can make it out... <gabber>that's a good question. maybe a recent software update changed emacs' ERC behaviour <Guest61>it does not seem to affect oftc though... <Guest61>mhm, I'll check my config again I guess... <jacereda>I'm running now on the broken config. Closing/reopening the lid didn't help. I removed /var/lib/gdm/.cache and the result is the same as before, black screen, I can see the cursor moving and that's all. I can switch to the terminal. <gabber>what happens if you press "ENTER" and then your password, then ENTER again (as if you would normally log in)? <gabber>is there anything interesting in /var/log/messages/gdm/greeter.log ? <jacereda>Typing the password seems to enter the session (I can tell because the fan starts spinning and the cursor changes), but the session is also black <gabber>have you tried roling back to your last working config? <gabber>what are the differences between the two? <jacereda>Rolling back works fine, the config didn't change, only the packages <jacereda>I'm on linux-lts with some tweaks to work on a macbook <jacereda>and the first time I noticed this was yesterday, I tried again today pulling the latest guix and the result is the same. <tschilptschilp23>A few weeks ago or so I also tried linux-lts and got some black screen issues, have you already tried the 'normal' kernel? <jacereda>Switching from linux-lts to linux didn't help, I'll try linux-libre now... <podiki[m]>trying some binaries via the FHS shell and hitting some "undefined symbol: _ZNSt18condition_variableD1Ev, version GLIBCXX_3.4.11" anyone have any ideas? some version mismatch somewhere? missing library?