IRC channel logs

2022-11-02.log

back to list of logs

<pkill9>since I updated my system, gnome-terminal got removed and console got installed
<Obikawa>jackhill: After some reading and more experimentation I have understood, that (package (arguments ...)) is on the boundary between the build host and the build target. I was able to break some stuff out into an internal let* expression. Thank you again for your helpful conversation.
<podiki[m]>is bordeaux not reachable for anyone else?
<ardon>podiki[m]: 🙋
<podiki[m]>ok, not just me then
<podiki[m]>main guix sub server is fine though
<civodul>podiki[m]: there's a hopefully short maintenance time for that one
<civodul>bordeaux.guix is back up!
***Xenguy_ is now known as Xenguy
<podiki[m]>Very short 😀
<user_>Hello
<user_>im getting this behavior during 'guix pull' so I cant update my system anymore
<user_> https://share.riseup.net/#f4Me6jZrq_f-Pg9_XzJ-yw
<user_>^ this is the output before it fails
<nira88>Is there a recommended way to configure the system bashrc?
***f1reflyylmao is now known as f1refly
<bdju>I had a syncthing daemon running on startup / all the time and somehow it's become screwed up and won't run on its own now. looks like I have a file at ~/.config/shepherd/init.d/syncthing.scm
<bdju>how am I supposed to start that?
<bdju>herd load root path-to-file gives me this: error: connect: /run/user/1000/shepherd/socket: Connection refused
<bdju>same with herd start
<cnx>how do i use local path for a package source
<cnx>context: i'm trying to make a manifest to build a program and drop into that shell
<cnx>hmmm don't manifest shell prepend to LIBRARY_PATH and C_INCLUDE_PATH?
<lilyp>cnx: only if you have gcc-toolchain in your manifest
<cnx>thanks
***Dynom_ is now known as Guest5004
***wielaard is now known as mjw
***jesopo is now known as jess
<ekaitz>hi! anyone has vscodium packaged for guix?
<ekaitz>(not that I want to use it, but I need it for reasons)
<nckx>[: The disappointing answer: https://paste.debian.net/plainh/d3d87d2d . No time travel was involved. The commit was made ~15 minutes after the release yesterday.
<nckx>'lo Guix.
<nckx>jacobk: Yes, all me. In retrospect the nick change caused a lot of confusion, pointless questions, and lost messages.
<nckx>Definitely doing it again next year.
<lechner>cnx: Can local files be a package source? I currently push to a Git repo and have a guix.scm that lags at least one commit behind. In the meantime, I try to test commits inside guix shell --development --file=guix.scm
<lilyp>lechner: yes, but be careful; guix will only see files checked into git if you use git-fetch for example
<lechner>lilyp: Hi, the idea was not to use git-fetch at all. Can I do that here, please? https://codeberg.org/lechner/guile-pam/src/commit/65fef0971761f5b70191350884429da5bddcb4fa/guix.scm#L50-L59
<lechner>Hi, is anyone using ZNC or pounce under Guix?
<lilyp>lechner: you'd have to replace the (origin ...) with a (local-file ...), but note that the git predicate is useful even if it's confusing
<nckx>lechner: Yes, but. There is no ZNC service and I haven't written one, if that's what you're hoping.
<lechner>lilyp: thanks! i understand the confusion part, but why is it useful please?
<lilyp>because .git can clobber your store with useless stuff
<lechner>and the git predicate ensures it will be ignored?
<lilyp>more than that, it makes sure only files known to git are used in the build
<tricon>\o
<nckx>.git is also not ‘reproducible’, in that it's very likely to differ between functionally identical checkouts.
<nckx>Hi tricon.
***civodul` is now known as civodul
<civodul>cbaines: shepherd on bayfront is slowly leaking :-/
<civodul>cbaines: you must experience the fibers-related leak in the Build Coordinator and Data Service, don't you?
<gnucode>hey guix, I am trying to write a simple guix shell command to launch firefox in a container that shares the ~/Downloads directory.
<gnucode>guix shell -CN --share=/home/joshua/Downloads firefox-wayland -- firefox
<gnucode>guix shell: error: mount: mount "/home/joshua" on "/tmp/guix-directory.lJP1TJ//home/joshua": Invalid argument
<gnucode>what am I doing wrong?
<rekado>is /home/joshua a bind mount?
<gnucode>rekado: I guess I'm not sure what a bind mount is...
<gnucode>mount | grep joshua
<gnucode>tmpfs on /home/joshua/tmp type tmpfs (rw,nosuid,nodev,relatime,size=3892492k)
<rekado>also useful to know: where are you running the command? By default “guix shell -C” will bind-mount the current working directory into the container.
<gnucode>rekado: I am trying to run that command in ~/
<gnucode>"/home/joshua"
<gnucode>so I cd into Downloads....
<gnucode>ran the same command
<gnucode>Error: no DISPLAY environment variable specified
<rekado>different error. Good.
<rekado>you won’t be able to easily run graphical applications in a container without sharing a lot more of your file system and preserving a bunch of variables
<gnucode>rekado: ahhhh. so this might be a bit of a chore to figure out.
<gnucode>hmmm
***civodul` is now known as civodul
<civodul>looks like IRC logging didn't survive bayfront reboot
<civodul>is that still done by a manually-started znc process? :-)
<rekado>possibly
<rekado>I wrote an IRC bot to replace this
<rekado>I was hoping for some comments on it before enabling it, though
<rekado>it’s in the maintenance repo
<civodul>ah neat, i think i've overlooked that
<nckx>Yeah, it's still ZNC.
<civodul>oh i remember now
<civodul>i'd say just got for it
<civodul>(it = goggles-bot)
<civodul>s/got/go/
<Kabouik>`guix gc` is "only" clearing 80GB of data after a `sudo guix system delete-generations 1m && guix package --delete-generations=1m`, but Guix still takes about 2/3 of my 1TB SSD. Is there a command I'm forgetting to clear space? I don't think just one-month-old generations alone would amount to that size
<nckx>Kabouik: There's also ‘guix pull --delete-generations=…’, and ~/.cache/guix/profiles/ (linked from /var/guix/gcroots/auto), and note that ‘guix package --delete-generations=1m’ without sudo clears only your user's generations, but that's… very, very much space. Almost implausibly much.
<nckx>I don't need to see the output, but take a look at /var/guix/gcroots to see if anything jumps out.
<nckx>I don't GC often and my store is ‘only’ 96GiB. My entire SSD is 500GB.
<Kabouik>That sounds a lot more reasonable indeed.
<Kabouik>My `guix gc` is still busy deleting the 80GB (it's very CPU intensive apparently), but I'll try a `guix pull --delete-generations=1m` afterwards, and a `sudo guix package […]` as well. ~/.cache/guix/profiles seems to contain only symlinks, I assume I shouldn't try to clean it manually.
<nckx>By convention, ~/.cache is always safe to delete (it may only cost time & bandwidth, of course, but never valuable data).
<nckx>That's not encouragement to delete random files, but I don't know the pretty UI commands to manage it (if they exist).
*nckx just deletes files in /var/guix when they run low on space, but that too is not legal advice.
<Kabouik>I'd say it is safe to play illegally when you're the game master, which you are.
<Kabouik>Now if I try to do that with my Guix system, I'm pretty sure I'm going to open a wormhole (and, coincidentally, break my system).
<nckx>Has more to do with old habits dying hard than mastery :) And yes, it is possible to break your system that way.
<apteryx>lilyp: is it expected that we have "+ out: everything else" even when only out exists?
<civodul>apteryx: i find it confusing too :-)
<nckx>apteryx: Yes, but +1 on confusing.
<nckx>(But then so is Appendix H, and I don't want to be a spoilsport.)
<lilyp>tbf the code is not very smart, it doesn't know that the list has only one element
<cbaines>civodul, talking about fibers leaking memory, I've spotted cases where the guix-build-coordinator memory usage just starts regularly doubling until it crashes
<lilyp>we could special case that to "everything" tho
<apteryx>and what is Appendix H supposed to be?
<cbaines>civodul, I don't think I've see the Guix Data Service behave like this though
<nckx>I'm not sure if it's quite a ‘joke’, but it's where all undocumented things are supposedly documented.
<nckx> https://en.wikipedia.org/wiki/Appendix_H
<lilyp>it's a nod to intel's appendix h, which contains undocumented opcodes
<nckx>
<apteryx>hehe
<apteryx>doc could probably say 'documentation' or something else than Appendix H ;-)
<nckx>Maybe me pointing out that it's a very GNU joke will trigger civodul into changing it.
<civodul>lilyp: wasn't the reference to "appendix H" removed?
<civodul>i thought there was consensus on it
<civodul>maybe i was the only one in consensus :-)
<nckx>Oh, I still have it, but I'm a few days behind.
<civodul>i think obscure geeky jokes are out of place; even i didn't get it
<nckx>(It's still there.)
<nckx>I think they're fine if they don't impede understanding. Here they do.
<civodul>yeah
<nckx>For some reason I thought you'd added it, civodul. Sorry for even thinking that.
<civodul> https://issues.guix.gnu.org/54454#15
<civodul>lilyp: how do you feel about posting patches for these two minor issues? :-)
<nckx>s/impede understanding/& and aren't exclusionary/
<lilyp>I wouldn't exactly say they're exclusionary, but then again perhaps I'm biased
<nckx>Nor do I, I was just being complete.
<nckx>Well, actually, one does kind of imply the other, but I certainly know it wasn't intended to be exclusionary.
<civodul>anyway, these two issues should be addressed
<civodul>cbaines: the leak could be slow; it seems to be a function of the number of context switches
<civodul>(imagine that in the kernel :-))
<nckx>Oof, is this spam? I really can't tell. https://issues.guix.gnu.org/58968
<nckx>Well, actually I do think it is, but it's probably best not to treat it as such just in case, is what I guess I think I mean.
<lilyp>It's certainly on the edge. Tell them that they've reached the bug tracker ;)
<Kabouik>Yeah, just finished a ncdu /gnu/store nckx, and it is using 719GB on my 1TB SSD. :| Meanwhile, `guix gc` is still busy.
<nckx>That is way beyond not even average.
<Kabouik>`guix pull --delete-generations=1m` deleted another 25GB, but that's not going to cut it
<nckx>The estimates given by ‘guix gc’ (really: the forked Nix code) are, if not flat-out wrong, correct in a way that is almost useless. However, I tend to see them over-, not underestimate savings…
<nckx>ncdu (1.17) seems to know about hard links, so that's not it.
<podiki[m]>did the logging (on logs.guix.gnu.org) stop sometime recently? seems to be from yesterday
<nckx>Yes.
<nckx>Kabouik: For me, the top consumers are image.isos, Linux checkouts, an indecent number of QEMUs, and of course everyone's favourite 1.5 GiB font, Google Noto. Is there anything notable about your ncdu output? Some forgotten 100-GiB disc images or so?
<nckx>(Disregard .links, that's deduplicated: https://paste.debian.net/plainh/839a527c)
<nckx>As you can see, this isn't some lean one-version-per-package installation.
<podiki[m]>TIL about ncdu, neat
<nckx>Big fan.
<podiki[m]>how does it handle things like btrfs compression? or is there a btrfs equivalent command?
<podiki[m]>(e.g. du isn't accurate on its own as far as i know)
<nckx> https://github.com/CyberShadow/btdu
<nckx>I've just heard of it, never used it.
<nckx>I have a WIP patch for it $omewhere, but that does not appear to be here.
<nckx>s/just/only/
<nckx>Though I have one server that still uses btrfs, the unconventional algorithm was really what prompted me to package it.
<podiki[m]>so you're saying I can't procrastinate later by packaging this, but you can by looking for the patch :)
<nckx>I wouldn't mind if you did, if that's what you mean.
<nckx>Warning: D.
<podiki[m]>my other work might
<podiki[m]>oh, never used something in D
<nckx>Neither had I, and I think our D build system was in a de facto broken state back then, and I also think someone has since improved (big ups to that person whose name I'll certainly never forget).
<nckx>*it.
<podiki[m]>I should fix https://issues.guix.gnu.org/58861 as a higher priority though
<podiki[m]>civodul: can you confirm that the commands in https://git.savannah.gnu.org/cgit/guix.git/commit/?id=c07b55eb94f8cfa9d0f56cfd97a16f2f7d842652 do produce an error? they didn't for me
<podiki[m]>civodul: though emulate-fhs does not work given a profile and I'm not sure how to handle that (say the profile wasn't made with emulate-fhs then it won't have glibc-for-fhs)
<nckx>podiki[m]: Commands?
<podiki[m]>err the guix shell invocation mentioned in the commit message
<nckx>You might not have linked what you meant to link.
<podiki[m]>oh wow whoops
<podiki[m]>and in the original message too, whoops
<podiki[m]> https://git.savannah.gnu.org/cgit/guix.git/commit/?id=8b192c5550213911f930594f4fd7386f36618237
<shcv[m]>how do I search for which package installs a given binary to the path? in this case I'm looking for lsblk (my system doesn't have it for some reason) and a basic search didn't return anything
<nckx>shcv[m]: util-linux (or util-linux-with-udev). There is no file search.
<civodul>podiki[m]: just replied! :-)
<civodul>thanks for the heads-up
<nckx>shcv[m]: Your system not having that by default is not impossible, but it's unusual.
<nckx>s/that/lsblk/
<shcv[m]>so the file search is "ask in #guix" then
<nckx>Correctomundo.
<nckx>Then someone runs realpath $(which FOO) and gives you the answer.
<nckx>:)
<jackhill>there are some old thread in the mailing lists brainstorming search ideas. The other thing I sometimes do, if figure out which Debian package has it, and look for the corresponding Guix package.
<podiki[m]>civodul: thanks! will try it out later
<nckx>You can request an ‘advanced search’ (ls /gnu/store/*-*/{s,}bin/FOO) but it costs extra.
<podiki[m]>though fhs won't work with a profile option to guix shell, any immediate thoughts on how to handle that? maybe warn that emuate-fhs may not work correctly if the profile was created without it?
<shcv[m]>@nckx: oh no, are we going to start a guix-coin? mined by build-server PoW
<podiki[m]>I also am guilty of using Arch's pacman/yay (aur helper) who owns a certain file and it tells me the package on arch at least, usually similar for guix (except a lot of smaller utility stuff sometimes)
<nckx>I should have added that the ‘advanced search’ is run on berlin.
<nckx>My money's on metaguixcoin, which gives out tokens each time someone suggests guixcoin.
<nckx>(It is tempting; I understand why.)
<shcv[m]>yeah, it doesn't really make sense for guix to integrate a blockchain (unless it was the guix version history itself), and external tokens are prone to all kinds of problems
<Kabouik>Nope nckx. I don't think I've got any .iso file in /gnu/store, or any virtual machine yet. I do have Noto, and would happily accuse it of eating 700GB but even that would not be fair to Google. My ncdu output just lists a 66GiB /.links, a 63GiB /trash (now deleted after guix gc I think), and then 5.3GiB hash-folders which I guess all contain the same share/ (5.3GiB) subfolder with links
<nckx>I can't parse what follows ‘5.3GiB’.
<nckx>These are directly under /gnu/store and don't have a meaningful name, just a hash?
<nckx>(Ignore whatever's under /gnu/store/.links, that's where we hide the implementation details.)
<nckx>shcv[m]: You had me at ‘doesn't make sense something something blockchain’.
<nckx>‘guix copy’ is unappreciated magic.
<Kabouik>Yes nckx, the 5.3GiB folders are those that are under /gnu/store and have a hash prefix. Actually I just noticed that they all are in the form of <somehash>-profile, I mistook them for application folders.
<Kabouik>I see now that I also have several <somehash>-qemu-7.1.0, not exactly sure how since I have not installed any VM (perhaps guix created them when trying I ran the suite of tests on some of the packages I submitted, I think there is a cross-compilation step), but that would only account for a few gigabytes anyway.
<nckx>Now that I find suspicious. Maybe there's some edge case I'm forgetting but profiles should not be several gigabytes in size.
<nckx>I mean, I'm sure I could artificially create one somehow, but look: https://paste.debian.net/plainh/4f8952cd
<Kabouik>All of it seems to be in their share/icons subfolder, and I installed Papirus, so that might not be default.
*nckx has to go AFK.
<Kabouik>Well first of all, just that short `du -hs` command takes ages to process on my machine, which I think is suspicious since you managed to run it on yours while we were chatting.
<Kabouik>Still not over.
<nckx>Kabouik: Back, but only briefly: don't cancel the du, but could you run ‘find /gnu/store/$hash-profile -type f -exec du -h {} + | sort -h’ (or whatever -exec command you prefer) on one of those gigantoprofiles? Profiles should be mostly symlink trees, so I'm expecting to see a generated (index) file that somehow gets 5.3 GiB of file data shoved into it. Somehow. But I don't really know what to expect.
<nckx>I'm on a 2013-era laptop with a 2016-era SSD, if that matters.
<nckx>Nothing beastly.
<Kabouik>I am just getting this: https://0x0.st/ogb2.txt But wait, ncdu has an "Apparent size" and a "Disk usage" mode, and hash-profile folders only show up with 5.3GB when in one of those modes (can't tell which, it's not indicated). I suspect one of those doesn't really process hardlinks the way we'd want?
<Kabouik>Yeah, my laptop should be more powerful, but du -hs on /gnu/store/*profile is still busy.
<Kabouik>And it is a SSD.
<Kabouik>I meant /gnu/store/*-profile
<Kabouik>Just finished: https://0x0.st/ogbp.txt
<Kabouik>It took half an hour apparently.
<nckx>Yes, ‘apparent size’ is exactly what we don't want. That's the naive ‘this thing is 1G, and that thing is 1G, so that's 2G used…’ mode that doesn't understand hard links.
<nckx>You *are* sure that /gnu/store is actually using ~700G, yeah?
<nckx>(E.g., and sorry, this will take twenty six days apparently, but: ‘du -hs /gnu/store’, no other options.)
*nckx AFK for good, or at least the rest of the evening. Happy hunting.
<Korven[m]>Anyone using Calibre from Guix, are you able to view your ebooks properly?
<lechner>Korven[m]: I recently opened an Ebook with Calibre but it compiled locally because no substitute was available. After reading about some security holes in Calibre, which is apparently setuid root, I switched to this https://depp.brause.cc/nov.el/
<Kabouik>Well I'm doubting it now nckx because I was trusting i3status-rust which reported only 373ish GiB of free space on my drive (I assume it uses `df` under the hood), and ncdu reported 718 GiB of "Total disk usage" (and 94 GiB of "apparent size", which is a lot less, but your comment implies that "apparent size" should overestimate, not the opposite)). `df -h` says I'm using 450GB and have 373GB free, which hardly adds up to my 950GB SSD, but this is
<Kabouik>perhaps because I am using btrfs and some space is reserved by the file system.
<Kabouik>I guess I'll run `du -hs /gnu/store` tomorrow and let you know the result for Xmas.
<Kabouik>Thanks for the help!
<nckx>Kabouik: It's possible I swapped them; I don't actually know ncdu's definition of ‘apparent size’, only du's, which is ‘apparent size > non-apparent (default) size’ in normal cases.
<Korven[m]>nov.el is nice, but what I meant was that if I view an ebook, it only shows the images in the book. No text. It works fine if I install Calibre from the Fedora repos, there's no problem
<nckx>(Yes, I *did* say that, but, my ride didn't show up yet, so here I still am.)
<Korven[m]>Uber?
<nckx>No, human.
<nckx>lechner: Calibre is not setuid root, cannot be, in Guix.
<Korven[m]>nckx: damn, wild
<nckx>Unless you are personally very aware.
<podiki[m]>is calibre setuid root in guix?
<nckx>No've course not.
<Kabouik>Perhaps it's the opposite then, yes, but in any case 94GiB doesn't add up either, my ~/ is less than 150GB, so there are still lost GBs somewhere if df -h only sees 373 free GB.
<Korven[m]>podiki[m]: do you use calibre too?
<nckx>I use ‘ncdu -qx /’ habitually, maybe try that first.
<nckx>(-q = don't waste cycles on a pretty animation; -x = don't cross mount boundaries, you really don't want to know how much imaginary stuff is in /proc.)
<Kabouik>It's not an emacs package, but you might be interested in shirah which was recently merged Korven[m].
<podiki[m]>Korven: occasionally, but not on my guix system I don't think
<nckx>(Korven[m]: It's just a friend, not a taxi, and she is habitually late.)
<Korven[m]>fair
<Korven[m]>Kabouik: Thanks, i'll go check it out rn
<Korven[m]>i really either just want calibre to work or another alternative to it :p
<Kabouik>Now that I think of it, I packaged it but don't really remember if it supports images.
<nckx>What format is the ebook?
<Kabouik>It might with sixel, but I wouldn't bet on it.
<nckx>I've used Guix's Calibre years ago and didn't notice such bugs.
<Korven[m]>nckx: just epubs and pdfs
<Korven[m]>If it matters, I'm on a foreign system
<Korven[m]>Funny thing is it can't even open the PDF files XD
<nckx>Hm, even the ‘Quick start guide’ it ships with is blank here.
<Korven[m]>D:
<nckx>Obviously because it's not setuid root. /os
<nckx>*/s
<nckx>Or /o\ works too.
<Korven[m]>hm?
<Korven[m]>i'm not sure I follow
<Kabouik>Shirah does support images, but opens it in an image viewer (`o` key) instead of showing them directly inline.
<nckx>Korven[m]: Nothing important. /o\ is our beloved protagonist from \o/ in a less cheerful mood.
<Korven[m]>ah
<Korven[m]>it also can't catalogue or edit the book metadat
<Korven[m]>s/metadat/metadata/
<Korven[m]>I might try emacs-calibredb, if nothing works
<Korven[m]>nckx: ah, that makes sense
<Korven[m]>i relate to /o\ as well, i'm in a less cheery mood
*nckx sends some good vibes over TCP.
<rekado>is anyone working on enabling logging again?
<rekado>or is that a task for future us?
<Kabouik>(Just for the sake of completeness, this shows Shirah's features: https://0x0.st/ogc6.mp4)
<nckx>rekado: I thought you were, above, so that's a no.
<Kabouik>Korven[m] ^
<nckx>I have no idea how to start goggles-bot.
<rekado>kinda busy now, but I can take a look later tonight
*civodul is still head-down on the shepherd leak, won't get their head out of it until it's done
<Korven[m]>> * <@nckx:libera.chat> sends some good vibes over TCP.
<Korven[m]>*packet received. sending confirmation frame*
*rekado also doesn’t know how to start goggles-bot
<civodul>(perhaps you won't hear from me ever again?)
<nckx>Didn't… didn't rekado write goggles-bot
<Korven[m]>I know I mixed like 5 different terms but, eh
<Korven[m]>Kabouik: tyvm
<rekado>yes, rekado did. But that was past rekado. This is the new me.
<rekado>I’ll do it in a few hours.
<rekado>civodul: don’t get lost!
<civodul>too late!
<Korven[m]>Kabouik: it got corrupted lmao
<Kabouik>The video?
<nckx>Ding dong. That's me. TTYL Kabouik.
<Korven[m]>yessir
<Kabouik>Safe travels nckx!
<Kabouik>The video just autoplays in Icecat for me Korven[m], but it plays in mpv as well
<rekado>civodul: leave a trail of bread crumbs.
<zimoun>civodul: hum, is it not the job of a sheperd to avoid to be lost? ;-)
<Korven[m]>hm, weird, maybe it's just on my end then
<Korven[m]>oh well, thanks for trying anyways
<two[m]>i
<two[m]>hi
<Kabouik>You should be able to wget https://0x0.st/ogc6.mp4 && mpv ogc6.mp4 I guess. Or just try Shirah, so that we double its user base among Guix users!
<two[m]>is it possible to build code for fhs systems?
<Kabouik>Got to go. Thanks for nov.el too, I didn't know about it and it looks nice.
<civodul>zimoun: indeed! and now i find myself looking after the shepherd
<zimoun>civodul: through Guix lens, everything looks like a bootstrap issue? :-)
<Korven[m]>Kabouik: laughed XD
<lechner>Korven[m] podiki[m] nckx: sorry, these articles were apparently outdated https://lwn.net/Articles/465311/ https://anarc.at/software/desktop/calibre/#problems-with-calibre
<podiki[m]>Ah I see, it was for mounting
<lispmacs[work]>pkill9: it was rather confused when I did my last upgrade, because gnome-console doesn't preserve the keyboard shortcuts
<lispmacs[work]>I can install gnome-terminal but it won't run, something about a missing .service file
<lispmacs[work]>now it is working
<apteryx>hmm, protobuf 3.21.9 fails to build with --system=i686-linux: https://paste.debian.net/1259282/
<apteryx>ah, seems reported: https://github.com/protocolbuffers/protobuf/issues/10418
<nashdidan[m]>My gnome installation does not seem to support drag and drop at all, even within the same app, and also does not persist preference for default apps. Does anyone else experience the same issues?
<klm_>If I want to package an AVR-specific project (https://github.com/dwtk/dwtk) that's written in go, should it go in avr.scm og golang.scm?
<ss2>Good evening. Is it a common problem that several gnome applications fail to launch while they cannot find libGLESv2.so.2?
<unmatched-paren>ss2: that's not normal, and i've never heard of it happening before
<unmatched-paren>is this in a python backtrace, by any chance?
<ss2>I'll prepare a paste.
<ss2> https://paste.rs/8l9
<ss2>that includes most of Gnome's core apps. Guix is current until yesterday. I've observed this since a week maybe. Just hadn't been testing gnome lately.
<unmatched-paren>hmm. the website isn't loading.
<unmatched-paren>could you please use paste.debian.net instead? or something else?
<ss2> https://paste.debian.net/1259286/
<unmatched-paren>thanks
<unmatched-paren>Hmm.
<unmatched-paren>Are you on a foreign distro?
<ss2>no.
<ss2>I checked with a vanilla user acount to rule out my profiles. It may have to do with my system.
*unmatched-paren guix shell gnome-text-editor
<unmatched-paren>it Works For Me
<ss2>okay. I better clean my system profile then. :)
<unmatched-paren>That's very odd, though.
<unmatched-paren>Could you try this first:?
<ss2>I guess you don't have (service gnome-desktop-service-type) in your profile?
<ss2>you mean guix shell gnome-text-editor? It fails there too.
<unmatched-paren>ss2: guix gc --referrers $(guix build gnome-text-editor)
<ss2>yes.
<unmatched-paren>and then the same with --requisites
<unmatched-paren>ss2: No, I don't.
<unmatched-paren>Though it might be good to try in guix shell --pure.
<ss2>--pure fails for other reasons. Will need more packages included to get it working.
<unmatched-paren>it shouldn't
<unmatched-paren>it *should* be possible to use --pure without any other packages and have it start fine
<unmatched-paren>that it doesn't indicates that there may be some dependencies missing or something
<ss2>first this: https://paste.debian.net/1259289/
<ss2>and output of gc --pure: https://paste.debian.net/1259290/
***mark_ is now known as mjw
<unmatched-paren>sorry, i'm not sure what's going on here...
<ss2>I may open an issue. Not sure what to do with it either.
<ss2>hm, gnome-text-editor runs on another machine.
***TopExpert is now known as snafuman
<two[m]><two[m]> "is it possible to build code for..." <- i did a `guix shell --container --emulate-fhs gcc-toolchain`
<two[m]>and ran gcc
<two[m]>but the resulting executable depends on /gnu
***daviid` is now known as daviid
<pippobaudo>hello everyone
<pippobaudo>I'm currently using debian 11 and I want to switch to guix because I want to experiment it
<pippobaudo>btw in the past I read that there were some efforts to make guix kernel agnostic
<pippobaudo>guix is it kernel agnostic?
<unmatched-paren>pippobaudo: it currently only works on hurd and linux
<unmatched-paren>there aren't any efforts for bsd afaik
<pippobaudo>isn't it developed in a way to make it work with any kernel?
<pippobaudo>like being easily portable
<pippobaudo>if I wanted to port it to bsd would that be easier than make let's say fedora works with bds?
<vivien>linux-libre*
<nckx>unmatched-paren: Is your 'two slashes' thing an in-joke?
<unmatched-paren>nckx: no? :)
<nckx>Ah.
<unmatched-paren>i'm just pointing out that the emoticon has an extra slash
<unmatched-paren>which presumably shouldn't be there...
<nckx>Escapement.
<unmatched-paren>Aha.
<unmatched-paren>/o\
<nckx>I didn't want to ruin the joke, if there was one, on the ML.
<nckx>Yeah, matching that in a substitute* would be "/o\\\\" :-)
<nckx>Which expresses it nicely.
<w0ll3q1uszxabiwo>How do you get the mirrored repository of inspekt3d from guix? The upstream is down and i want to fork it
<w0ll3q1uszxabiwo>and guix seems to be the only source which has the package
<nckx>Which part of ‘ban evasion is frowned upon’ was unclear?
***ChanServ sets mode: +o litharge
***litharge sets mode: +b $a:KREYREN
***w0ll3q1uszxabiwo was kicked by litharge (You are banned from this channel (by nckx))
***litharge sets mode: -o litharge
<nckx>(It was a rhetorical question.)
<user_>Hello
<user_>there is a Way to reinstall guix to fix integrity?
<nckx>What is damaged?
<nckx>(& hello!)
<user_>Im not able to guix pull anymore let me paste the output
<vivien>Not here please :)
<nckx>Alright! (Just in case: not here or the bot'll get you; ideally use the pastebin URL in the topic.)
<user_>I will upload it to share.riseup.net (I always use this one for pastes)
<anschlussy>Do the guix developers/creators hang out in here?
<vagrantc>sure do
<anschlussy>That's cool. I like to see people active in the places where they can be most easily contacted.
<user_> https://share.riseup.net/#XDo_fsMmFOpzxJNWQrbx7A
<user_>this is the output of guix pull
<anschlussy>Before I start whining, I hope everyone feels appreciated for working on a modern system dedicated to free software. I thought that kind of thing was dead.
<unmatched-paren>user_: "JavaScript is required to view and upload files" :/
<anschlussy>Is that error completely empty? lol
<anschlussy>unmatched-paren: I can re-upload somewhere
<unmatched-paren>please use something like paste.debian.net instead
<anschlussy> https://paste.debian.net/1259302/
<anschlussy>There it is; that's user_'s paste
<unmatched-paren>thanks
<unmatched-paren>oh, uh
<user_>Will tryt paste.debian.net sec
<anschlussy>user_: don't worry, I got it.
<lechner>Hi, does anyone using 'pass' have completion for the site part holding the scecret in either Bash or Fish?
<unmatched-paren>shot in the dark, user_, but maybe try sudo guix gc --verify=repair,contents
<user_> blob:https://share.riseup.net/6f91ba82-84a5-49f9-9117-95508998e75f
<anschlussy>I've wanted to try out pass, lechner; wouldn't that usually be in some sort of "bash-completions" package?
<user_>^ just to test this way also need js?
<anschlussy>blob URLs are weird, user_. they're not real urls, so it's referring to something that's already been loaded, if i remember correctly
<vivien>That error message looks scary
<unmatched-paren>user3456: "404 page not found"
<anschlussy>vivien: interesting how the error is all 0s, lmao
<lechner>anschlussy: pass may ship its own shell snippet, but i also have bash-completion installed (via Home)
<user_> https://paste.debian.net/1259303/
<anschlussy>Also, while I'm here, it looks like people have this whole "nonguix" kernel on some of their systems, for laptops and the like. Would I need something like nonguix in order to make use of amdgpupro for GPU computing stuff?
<user_>this paste.debian is very fine I will start to use this one from now
<user_>no js and also dont blocks tor
<vivien>user_, the log mentions /gnu/store/cpc4zc2cn4sslr043lxf5virg7p…, could you try and find that file to see if it has the zeros?
<anschlussy>it looks like donenfield wrote the .bash-completion himself, lechner, thinking ahead: https://github.com/zx2c4/password-store/blob/master/src/completion/pass.bash-completion so it should be part of any package you got. I'm not on guix, so you should feel free to ignore me--I'm not very versed in anything.
<user_>Let me chec here
<vivien>(you need to have the shell auto-complete the part before the dots)
<anschlussy>I've also gotta head out, but I'll be back in here later.
<unmatched-paren>anschlussy: unfortunately GNU policy states that you must not help with or drive people towards nonfree software on official channels
<unmatched-paren>so you'll need to head to (that channel)'s IRC... channel.
<anschlussy>lmao, i see. I'd also like to put in that i'd rather throttle the CEO of amd until they released the source for the prop. drivers
<anschlussy>or maybe someone could come around and hack the pants off of them like they did to nvidia
<lechner>anschlussy: well, it's not working
<anschlussy>lechner: is that file on your system somewhere?
<user_>vivien: the folder is "/gnu/store/cpc4zc2cn4sslr043lxf5virg7pamj1n-guix-e26b529" but Idk which file caused the problem
<anschlussy>something like "find / -iname 'pass.bash-completion'" as root would determine if it's there, right
<vivien>Of course :/
<vivien>Sorry I should have known that
<user_>I should run this find?
<anschlussy>user_: nope, i was talking to lechner
<user_>Oh ok
<unmatched-paren>anschlussy: probably more efficient to do find $(guix build password-store) -iname pass.bash-completion
<anschlussy>unmatched-paren: guix rules
<anschlussy>that's the kind of thing i like to see
<unmatched-paren>(guix build prints the store path of the output when it's done building)
<unmatched-paren>and if there's a substitute, it won't build it at all, just download and print
<unmatched-paren>and if the substitute is already in the store, it'll just print :P
<vivien>user_, I’m a bit clueless, maybe it’s a cache issue? If you ran guix pull as a user, then maybe remove ~/.cache/guix and retry
<nckx>user_: Thanks! Sorry, I was distracted. Have you reported this issue via mail? Somebody has, I've seen this exact thing today, but frustratingly I'm not finding it now.
<nckx>But maybe it was you.
<user_>I postd something few days ago but here on channel
<user_>I still not reported via email (I should do this)
<user_>vivien: Im running the guix gc verify
<user_>I will check the cache later too
***bjc` is now known as bjc
<nckx>OK, it was zi​moun's reply I saw today, the actual bug is older, but suspiciously similar: https://issues.guix.gnu.org/58802
<nckx>--verify=contents,repair is an excellent first step though.
<lechner>unmatched-paren anschlussy: thanks for the command. it applied a bunch of grafts the first time around. then it came up empty
<lechner>unmatched-paren anschlussy: a modified command found /gnu/store/axn4azbkcd6zxbjiza8wjk0vd51gq0n5-password-store-1.7.4/etc/bash_completion.d
<user_>Did u guys tried guix with hurd?
<nckx>I played around with it a bit back when it was first available, but nothing more than play.
<nikolar>Hurd isn't exactly production ready
<rekado>depends on what you produce
<nckx>Nice.
<user_>the debian/hurd seen to have lot of things working fine even x11
<user_>nikolar: there are people working on hurd itself acctually?
<nikolar>Yeah, there are *some* people working on hurd
<civodul>hi goggles-bot!
<nckx>sneek: later tell goggles-bot: botsnack.
<sneek>:)
*vagrantc tries reviewing ~270 guix lint issues around "(description|synopsis) should start with an upper-case letter or digit" and wonders if the check should just be removed
<nckx>Assuming goggles-bot uses the same file/format as ZNC, I'll copy the missing log data when the bot's no longer writing to today's, and if that's all right with everyone.
<vagrantc>although I've found at least a couple feature requests for guix lint to improve
<nckx>vagrantc: Hm, I wonder, does it handle a leading @ at all?
<nckx>That would lead to a lot of false warnings otherwise.
<nikolar>user_: ^
<vagrantc>nckx: yeah, there are some issues around @code ... both with the upper-case and i think in some cases with line-length
<vagrantc>e.g. @code{} is counted in the line-length
<vagrantc>and there's a limit of 80 characters ... but i think it renders without those characters
<nckx>That sounds correct though, if I read you right.
<nckx>The line length is for the code; not the rendered output.
<nckx>But maybe I don't.
<rekado>did sneek just snatch the snack for goggles-bot? ts ts
<rekado>nckx: yes, same format as ZNC
<vagrantc>(if (>= (string-length synopsis) 80) ... should become (if (>= (string-length (substitute synopsis ("@code{" "")) 80) ... or does substitute only work on files?
<vagrantc>and would also need to strip out the trailing }
<nckx>Excellent.
*vagrantc grasps at cargo-culted straws
<vagrantc>more likely i want a string-SOMETHING
<nckx>There are much simpler built-ins than substitute* for that, yes :)
<nckx>I assume you meant substitute*, whose name implies the existence of a non-prime substitute, which I've literally never used.
<vagrantc>string-make-string-go-away
<vagrantc>string-drop ?
<two[m]>string-length is not accurate btw
<two[m]>(string-length "я́") = 2, (string-length "🧛‍♀️️") = 5
<vagrantc>string-delete ? though i can't parse the explanation of how to use it https://www.gnu.org/software/guile/manual/html_node/Miscellaneous-String-Operations.html
<two[m]>perhaps counting words would be better?
<vagrantc>two[m]: that's at least an order of magnitude more complicated than i'm ready for ... i can even figure out how to remove a string.
<vagrantc>remove something from a string
<vagrantc>but i can string-append with the mediocreist of em
<nckx>vagrantc: I just meant string-replace-substring, nothing fancy.
<nckx>But you can get fancy & use regexps if you want.
<vagrantc>fancy is not my style
<nckx>two[m]: я́ is fair, but I think punting on emoji modifiers (‘vampire + female’, not even my modern terminal renders that as one glyph) is… reasonable, or at least not intolerable.
<vagrantc>yay!
<two[m]>(define words (make-regexp "[[:alpha:]]+" regexp/extended))