<rekado>pkill9: override the gnome package in the gnome service type with a modified package that doesn’t propagate all of the gnome packages. <pkill9>is there anything packaged in guix that allows you to stream files within torrents? <pkill9>rekado: are you able to set default web browser to ungoogled-chromium? I can't seem to do it <pkill9>doesn't give me any other option in gnome-shell <Kabouik>My vainfo is fixed indeed nckhexen, thanks; I don't think I could have found by myself that mesa is installed in the system profile and not i-m-d and libva <Kabouik>Sorry to answer only now but it needed a reboot before showing up as fixed <tschilptschilp23>Hi guix! I just updated from e05e5342a0baa8a72d3e1b1e3a5d57da13f764b3 to bb3b810167d9ff784770a848a6f86b0cfa9cdfd8 and found some gnome related things to apparently be broken now. 1) Evolution does not recognize my configuration anymore (mail, calendar full-on affected). 2) Secrets (the new passwordsafe) does not open, nor create kdbx files anymore. 3) Gnome-maps can only be opened from a terminal anymore, the launcher does not work. <vagrantc>tschilptschilp23: have you logged out and back in again? <vagrantc>gnome relies on some environment variables to be set up, i think <pkill9>is there a way to give the option in gdm to run X gnome or wayland Gnome? <clever>pkill9: last i looked, gdb is just an X client, so the X must already be running (or xwayland?), ive not looked into wayland much yet <pkill9>nevermind, it already gives me the option <podiki[m]>hilarious, working on guix container and somehow I guess removed the user name, so the bash prompt is "I have no name!" instead of a user <clever>podiki[m]: does /etc/passwd exist? <podiki[m]>clever: seems to happen when using the --bootstrap option; yes /etc/passwd exists <attila_lendvai>i wish there was a way to avoid updating packages that have no substitutes... there's no way i want to build firefox locally right now. i wish i could just be done with a home reconfigure and see if it fixes nautilus. <Schmoho>I am having an unnerving issue using guix as an auxilliary package manager in Mint - I've mostly been using it to install emacs-native-comp packages and this apparently depends on a glibc-2.33 version. <Schmoho>Now there is a package in the default repos on Mint that I need to install, which after installation fails to start, telling me <Schmoho> symbol lookup error: /gnu/store/5h2w4qi9hk1qzzgi1w83220ydslinr4s-glibc-2.33/lib/libpthread.so.0: undefined symbol: __libc_pthread_init, version GLIBC_PRIVATE <Schmoho>Shouldn't it be impossible for a package installed via apt to be linking against a guix-store libc of all packages? <efraim>Schmoho: do you have LD_PRELOAD set anywhere? That often is the problem <Schmoho>efraim: I really don't think so. Got to admit, this was the first time I heard of it. Couldn't find it in any shell configs and its not set in env <nashdidan[m]><attila_lendvai> "FTR, this is the nautilus issue..." <- I also reported it, though didn’t get a reply. Also noticed the the sub commands of tracker3 itself don’t exist, as mentioned int the thread. Didn’t try yet but does installing tracker@3 work? <attila_lendvai>nashdidan[m], i didn't try either, but on the issue it's reported not to work. <nashdidan[m]>attila_lendvai: Ok, thanks. Indeed it doens't work, though your workaround does! <numerobis>Hi #guix. Since my last update of guix, I have an issue with gitolite. It seems that when I create a new repository, the "hook" folder is not present, which leads to an error. See https://paste.debian.net/1255805 Has anyone else experienced this issue? <civodul>numerobis: hi! what's in /var/lib/gitolite/.gitolite/hooks/common ? <numerobis>civodul: there is one default hook called "update" <numerobis>Woops, no sorry, I misremembered. The folders contains two folders: "common", which contains "update", and "gitolite-admin", which contains "post-update". <numerobis>In fact my first answer was correct, sorry for the confusion. <cbaines>numerobis, how are you creating the new repository? <numerobis>The file /var/lib/gitolite/.gitolite/hooks/common does exist, so i wonder what could cause the symlinking to fail. <numerobis>cbaines: I `git init` and `git commit` on my laptop, then `git remote add origin gitolite@hostname:reponame`, then `git push`. When I check in /var/lib/gitolite/repositories, I see that the repository does get created, but the hooks subfolder in .git is missing. <cbaines>numerobis, so you're not interacting with the gitolite-admin repository? <numerobis>got to go but I'll reconnect later. Thank you civodul and cbaines! <sektor[m]>This is not how I wanted to start the morning off... ***wielaard is now known as mjw
<abrenon>I'm trying to install a piece of software with outdated dependencies. I may have located a commit in guix' history when the required versions are all still present <abrenon>I'm using guix time-machine to try and get an environment where the dependencies are available <abrenon>would that work to actually build the package from a proper guix definition (as in, a .scm file) or would I need to understand inferiors and such to achieve this ? <sektor[m]>I'm honestly not sure; I am re-getting started with Guix TBH. <abrenon>thanks ! Don't worry I'm still reading the doc about inferiors again, but just thought I'd asked in case anyone had some clear insight <sektor[m]>I'm wondering if when you inherit from the installation os and modify things if it makes a separate boot menu. I don't see any of the packages I put into it. <sektor[m]>That and I don't think it took my kernel changes, either. <abrenon>that's suspicious: it most probably didn't work at all <abrenon>AFAIU writing the declaration of a system by inheriting from another one shouldn't have any impact on the grub boot menus <abrenon>the separate entries should only appear as you create new generations of your system <abrenon>so if you've built a new generation with this new declaration inheriting the install media, then it should've indeed created a new entry, but it should be the one by default in any case <abrenon>have you done `guix system reconfigure` and not only `guix system build` ? <sektor[m]>Curious. I should put the file somewhere. I'm pretty confused. <sektor[m]>This is virtual; I'm bootstrapping msyelf a bit. <abrenon>more complex situations are always good reasons to forget easy things and hit friction <abrenon>so please tell us more about your process, maybe you'll realize something obvious as you do : ) <sektor[m]>Hopefuly once I've got these ironed out we can get them into Guix proper. <sektor[m]>Naturally, there's the issue of getting audio working and writing the actually espeakup service, but gotta walk before running. <abrenon>ok, I've never compiled my kernel myself but this looks good <abrenon>I suppose the truncated begining is a (define %linux-a11y-options ? <abrenon>oh, so you're actually building an installer, not configuring your system <abrenon>ok so you have this system definition you showed me, and you successfully got an .iso out of it ? <abrenon>when this iso is booted, it only has one menu entry, which should be the one for your new system <abrenon>the system successfully boots (again, for some definition of "successfully") <abrenon>but it's nothing like the system described in the definition ? <sektor[m]>No. I tried which w3m, got command not found. Also the speakup and alsamixer programs are missing. <abrenon>ok so probably a mistake somewhere else I guess ? another iso got burnt ? physical media has some weird setup which still uses an old boot config from a different partition where the regular installer happens to be located ? <abrenon>what system are you using guix from ? <abrenon>(as in, the command line eating the system declaration) <sektor[m]>Arch right now; I'm booting them in Libvirt now. <abrenon>no idea what could be going wrong but I still think it must be something relatively obvious <sektor[m]>Agreed; the trick is to figure out where I went wrong. <sektor[m]>Incidentaly, is there a way to have qemu write to the terminal? I noticed tye system vm optin, but after a time the screenreader on the host only said "terminal". <abrenon>you could mount the iso and look directly in the filesystem to check the obvious : additional packages are there <abrenon>sorry I don't remember precisely but I've stumbled on that a couple days ago <abrenon>there's a way to ask for qemu's -nographic option <abrenon>but I don't remember if it's a default or something with another option <abrenon>anyone knows what py2app is or how it works ? I've got this application I'm trying to package which setup.py says to run python setup.py py2app <abrenon>I can run it directly when entering a container with the right dependencies (it breaks further down the line but that's another matter) <abrenon>but when I try to launch the build autonomously with guix build, the environment seems to think it's missing a py2app pip package and trying to retrieve that from the network <abrenon>what I don't understand is that it looks like a MacOS tool so I don't understand why it'd be required to compile the software I want for linux <abrenon>what I even less understand is why the installation script doesn't seem to be trying to install it from the container, as if it was somehow already available ?! <abrenon>which is why I wanted to ask if anyone had any prior knowledge of it and how it's used <sektor[m]>Maybe the thing wasn't cross-platformed properly, and had a Mac-first design? <abrenon>I'm afraid it must be something like that : ( <sektor[m]>Interesting, I should put some of the logs here from Qemu somewhere; I think it hit something fatal. <abrenon>sektor[m]: has the --no-graphic option been of any use ? <sektor[m]>It was mainly so I can read what's going on. <sektor[m]>I need to do this in eshell though to read the log in it entirity. <apteryx>ocaml-craml is broken, in case some ocaml person would like to fix it up :-) <civodul>i wonder how many people were left behind like this <attila_lendvai>civodul, i've done quite some thinking about the same issue, but with self-hosting programming languages. there's a chain of evolutionary stages of the language that can bootstrap each other in a row, but skipping one is not possible. i've decided to make these explicit in consequtively named git branches. my experiments are documented here: https://github.com/attila-lendvai/maru/blob/maru.10/doc/bootstrap.md <attila_lendvai>in parallel (i think) a guix pull can only pull up to so many commits before it needs to build and replace itself to pull further <apteryx>civodul: not ideal for sure, but I'm not surprise upgrading from guix-daemon (GNU Guix) 1.0.1 doesn't work <pkill9>is there a tool for both reading and sending to mailing lists? <apteryx>perhaps we should add a --re-install option to our guix-install.sh script that would be handy in these cases <apteryx>or simply prompt: 'GNU Guix installation already detected; overwrite it? (y/N)' <antipode>pkill9: For _reading_, there is public-inbox and your usual e-mail program. <antipode>For writing, there is "git send-email" and your usual e-mail program <antipode>Other tools might exist as well but I don't know of them ***meonormativity is now known as drmeo
<rekado>civodul: I recently upgraded an ancient Guix installation. <rekado>the easiest way to do that was *not* to use “guix pull” with intermediate commits, but to substitute a recent Guix via derivation <rekado>I went to ci.guix.gnu.org, looked at a recent build of the “guix” jobset, and substituted the derivation. <rekado>then started the daemon from there and used the guix from there to run “guix pull” <rekado>in my experience that’s much faster and much more convenient than to hop from commit to commit. <dthompson>rekado: hey that's a neat trick! every now and then I try to update an ancient installation and it's certainly tricky <pkal>I am trying to build an installation image using "guix system image", but this error keeps occuring: https://0x0.st/oJZr.txt. It says "No space left on device", but df -h /gnu/ keeys saying I have over 300 GB left. <mothacehe>the image is built in /tmp so 1.6 GB may not be enough <pkal>Is there any way to say "use the current working directory"? I couldn't find anything in the manual from a quick search <mohamed>I can't install the "rust-cargo" package, guix home is telling me that he can't find the name even when I add "gnu packages crates-io" <pkal>mothacehe: Sadly that didn't seem to work, or at least it seems like I have to restart the entire daemon with a custom TMPDIR <unwox>mohamed: try rust-cargo-0.53 <mothacehe>yes restarting is required, can't you bind mount /tmp to some disk backed directory? <civodul>rekado: clever! problem is that the recent guix won't have gzipped substitutes, so that wouldn't work in the case above <civodul>attila_lendvai: that's inspiring; so far we manage for 'guix pull' to not have an intermediate step, by assuming a small subset of Guile/Guix APIs will "always" be available <civodul>but maybe at some point we'll need to do what you describe <pkal>mothacehe: Looks like that worked, thanks for the tip! <mothacehe>pkal: great :) we should probably add a daemon warning when /tmp is full/too small <tricon>the Guix logo and the "kaleidescope" aesthetic for the 10-year are so, so good. kudos to the designers. <pkal>^^ Or an option to reconfigure what output directory to use... <pkal>Especially when like in the case of "guix system image" the output of the command is just a single file. <nckhexen>Guix always stores (heh) output in the store. <pkal>nckhexen: I just realised that the output file wasn't in TMPDIR, but under /gnu/. My bad. <apteryx>mothacehe: hi! looking at pending builds at https://ci.guix.gnu.org/metrics, there are more than 50k at teh moment, and this number seems to stay around 40k even when the build farm is idling; what is this symptomatic of? <nckhexen>Are you sure aarch64 ever idles? I've seen offline & busy, never idle 😛 <apteryx>nckhexen: kreuzberg is forever idling <rekado>do you have access to all of the aarch64 nodes? <rekado>if there’s a need to visit the data centre to take a look at them, please let me know and I’ll see when I can schedule it. <civodul>rekado: most likely kreuzberg is building but ci.guix is unable to retrieve substitutes from there <civodul>like i wrote on guix-sysadmin, i'm unable to access it and fix it like the two other boxes <civodul>well, "boxes" is a strong word i guess :-) <rekado>ok, I’ll be sure to check on it when I’ll visit there again <rekado>(I’m hamstrung by the hinge failure of my Librem 13. The front fell off.) <masot>(whoops sorry, wrong window :p) <civodul>rekado: if you can log in, perhaps you don't need to physically access it <civodul>i get "You are required to change your password immediately" when attempting to log in <civodul>apteryx: dover was off for a while and recently moved to cbaines's place <apteryx>seems overdrive1 and dover are missing from machines.rec <civodul>(which, given the name, makes a lot of sense) <civodul>i think machines.rec suffers from bitrot <civodul>problem is at some point machines.rec became almost-but-not-quite redundant with machines-for-berlin.scm <civodul>we should consolidate that, and prolly use Scheme instead of recutils for the additional metadata? <pkill9>rekado: do you run gnome with xorg or wayland? <rekado>I tried wayland, but there were some very annoying quirks that I wasn’t able to work around (I don’t remember the details), so I’m back to xorg. <apteryx>civodul: not all the machines in the Guix build farms, collectively owned by the project are connected by berlin, so this separate file still has value, I'd think <apteryx>rekado: are you able to ping berlin from it, using its wireguard IP? <apteryx>pinging from berlin works (10.0.0.9) <civodul>i can log in on kreuzberg now so i'll try and reconfigure it like pankow and grunewald <f3n1x>i need to print libreoffice document as a pdf. I'm getting an input/output error will saving it after editing. What am i missing ? Thanks, thanks, thanks <f3n1x>Ah tricon, ok. i mean, the pdf is generated. The .odt file saving operation is what gives the error, thus leaving libreoffice unusable in my machine. *f3n1x is gonna try with abiword <f3n1x>the problem ^^ is not reproduced while using abiword ! <antipode>civodul, attila_lendvai: A mechanism for automating intermediate steps would also be useful for incompatible changes to the channel authentication mechanism (e.g. new key types, support for PR-instead-of-rebase, ...) <antipode>mohamed: You probably need rust:cargo instead (i.e., the 'cargo' output of 'rust') (IIUC, rust-cargo is for if you use cargo as a library in a Rust program). <nckhexen>I hadn't noticed that dmitri/sergei were commented out. They are still powered on & consuming power AFAIK. <apteryx>rekado: if wireguard pings OK from kreuzberg to berlin, I'd then try restarting cuirass-remote-worker <nckhexen>I just meant ‘not in the /workers list’. <nckhexen>I didn't look into it, just noticed it now. <rekado>apteryx: yes, I can ping 10.0.0.1 just fine <rekado>restarting the cuirass-remote-worker now <nckhexen>Not a big deal, but they should be able to be added back. <apteryx>OK! then perhaps the worker just needs to be kicked <f3n1x>usually i use ufw for easily managing firewalling, adding iptables rules etc... I see there is no ufw package available in guix. Aside from managing iptables manually, are there other easie alternatives i could use in guix ? ***Dynom_ is now known as Guest6660
<minima>hi, i don't seem to be able to have bash completion with pass (password-store) - i.e. my pass doesn't tab-complete my passwords <rekado>hmm, kreuzberg is now listed on /workers, but pankow is not <minima>anyone else experiencing something similar? my pass and my bash completion are otherwise fine, apparently <minima>(there doesn't seem to be trace of pass under .guix-home/profile/share/bash-completion/completions/ as a matter of fact) <minima>oh... however... i do see pass under '.guix-home/profile/etc/bash_completion.d/' <viaken>Anyone else done an install recently with an encrypted btrfs /? I can't seem to get it to work. <viaken>GRUB is throwing errors trying to unencrypt the partition. <antipode>viaken: Did you use the graphical installer? <antipode>If you do a manual install, there is a risk of using a 'wrong' encryption type that GRUB doesn't support yet. <antipode>(a manual install is fine too, as long as you use the right types.) <viaken>I did use the graphical installer, yes. I'd attempted manual before and saw that only LUKS1 is supported. <viaken>Oh hold on, let me check if it's a keymap issue. <antipode>(On 'I did use the graphical installer ...'): Then I don't know, it would be helpful to share the actual throwed errors. <viaken>Keymap issue. It seems it uses US. <viaken>I'm logged in. Thank you for the help. ***mark_ is now known as mjw
<Grimpper>Hi. I'm trying to package the simplest thing ever but I get an error un the install phase. This is the package, it is just a c file that get's compiled into an elf: <Grimpper>This is the error: `install: cannot create regular file '/gnu/store/llkjb4hr6y3zdfpphlzl1rq067ydbqwg-xwinwrap-0.0.6/bin': No such file or directory` <Grimpper>seems that it would be a pretty easy thing to fix but I'm to noob to figure it out :'( <itd_>Grimpper: one idea would be to teach `install` (see it's options) to create the `bin` directory (which seems to missing, if I'm not mistaken) <csepp>hey, i'm trying to package some ocaml libraries and could use some help. <csepp>specifically i'm building emile and it wants ipaddr, which i am passing as an input, but it still doesn't find it. <apteryx>anowhat about pankow? it's not listed on the workers page <apteryx>ah, just saw you hard already noticed the above, rekado <viaken>Is there a reason one would prefer (append (list foo) barlist) over (cons* foo barlist)? <apteryx>no, except it's supposed to be more readable <apteryx>(in your later example you don't need cons*, you could use just cons since there's a single entry to prepend <viaken>(Right, I just skipped other potential entries.)) <sektor[m]>Logs from the most recent boot; sorry to keep you waiting, I had gotten distracted. ***maximed is now known as antipode
<yewscion>Could someone help me understand something about the GWL? I'm not sure where the output is going when I run `guix workflow run`; It seems that the processes are being run properly, but no output other than the info is displayed to the screen. <shcv[m]>what do people like to use for an MTA on guix these days? <old>yewscion: I can help. Outputs are supposed to be written in the working directory where you've run the workflow command. Except when it's running in a container. In that case, it's copied from the container into your working directory afterwards I believe. <old>yewscion: You might want to ask the question on #guix-hpc however. You have better chance for help there <mekeor[m]>how can i find out why i have this package on my disk or which package i have to remove so that this package gets removed after a gc? /gnu/store/qlcn7vp0jj3wl6dljf5wxpk5ygygsl7z-texlive-texmf-20210325/ <rekado>apteryx: “no route to host” for all but kreuzberg ¯\_(ツ)_/¯ <yewscion>old: Copy that, thanks for the advice. I've been trying to figure out where the output has been going for a while now; Should have asked earlier. I'll hop on #guix-hpc now to keep conversations in their most-relevant places. <apteryx>rekado: OK! I guess these will benefit from a reset when you visit at the MDC <apteryx>I'm struggling to help a user whose /etc/profile.d/guix.sh is not sourced by their Ubuntu 20.04's GDM for some reason <apteryx>it appears to be in a test VM, so I'm puzzled