IRC channel logs
2025-01-27.log
back to list of logs
<Deltafire>anyone know which package has the 'reset' command? <graywolf>How do I do guix shell -C with forwarding for network? <meaty>What should I do when a package wants to use a plain makefile and pkg-config to build? It keeps looking for .pc files that aren't packaged or don't exist in the build env <meaty>or does anyone know of another package that does that? no autotools/config, just Makefiles. I'm trying to update tuxpaint <graywolf>jakef: thx; I even am using that option it other scripts... <jakef>in a build log I'm getting: patch-shebang: *.py: warning: no binary for interpreter `python' found in $PATH <jakef>i thought it would look for python3 too? <jakef>i guess using python-wrapper instead of python would make it work <coyotes4ys>i am on an 8.9 inch display, need slightly bigger font for generally everything but not sure if i want to change the resolution due to screen space <coyotes4ys>xrandr i know of but not sure exactly what to do or how to fig out what to do <coyotes4ys>where are the openbox config files? nothing in my .config folder and this openbox walkthrough says /etc/xdg which does not exist <graywolf>coyotes4ys: probably $(guix build openbox)/etc/xdg/openbox <graywolf>Sorry I am not sure what you mean. You do not have etc/xdg in the openbox package? <graywolf>Can you paste somewhere output of `find $(guix build openbox)' and share the link? <coyotes4ys>oh it's not on my drive except in the guix system package? the whole os is a package because i installed guix system with no other os? <graywolf>Guix packages in general do not propagate their parts into /etc. Guix services (sometimes) do. In this case you probably should copy the configuration from the guix package (the path $(guix build openbox)/etc/xdg/openbox I pasted above) into your home. <graywolf>(Ideally there would openbox-service-type, I do not know whether there is.) <coyotes4ys>i am new to this guix stuff. i half understand and half don't <coyotes4ys>anyway the library's closing i need to go elsewhere and i may find you then <graywolf>No worries, it was hard for me at the start as well. (And it still is hard.) <coyotes4ys>is there a page in the tutorial on the guix site that helps with understanding where this stuff i want is and how to copy it outta the package onto my home? <sleepylkw>any ways of speeding up guix pull running on a foreign distro? <apteryx>fun fact: to apply patches containing patches via Gnus with 'gnus-summary-pipe-output', you need to specify the M-i r "symbol" prefixes, meaning raw mode (to keep all headers). If you need to apply e.g. 4 patches, you'd do 'C-u 4 M-i r |' on the first message. <meaty>could somebody verify that it's not just my system? <meaty>also the current version is broken, as it autodetects the latest SDL version installed and refuses to launch cos it wants an older version, even in 'guix shell --container --pure vkquake sdl2@2.0' <wakyct>what error does it crash with meaty? <podiki>note that your container command above doesn't change anything about the sdl2 vkquake uses, as that is in its package definition <podiki>if you want it build against an older version, you need to specify that in the inputs it uses <podiki>e.g. use the variable sdl2-2.0 instead of sdl2 <meaty>podiki: That's using the "current" vkquake package <podiki>i just went based on the paste you had above with a package definition <meaty>yes, that's the new one I'm writing <podiki>my comment about the container stands though, that won't change what vkquake is linked against/opening <meaty>the one currently in guix is broken, then <podiki>so yeah, you want to try building vkquake with sdl2-2.0 if that's what it wants <podiki>trying "guix shell vkquake --with-input=sdl2=sdl2@2.0 -- vkquake" complains it can't load gfx.wad as i don't have any files i was giving it (i've never tried vkquake) <podiki>haven't had quake files in...way too many years :) <podiki>seeing as the guix version is many years out of date, might take some doing to update it to current version <sleepylkw>hey my guix pull is stuck at 80% and has been for a few hours, anyone hit this? <adanska>theres a problem with the boost sources, trying to build from source gives the following error: https://paste.debian.net/1346995/. there's now a hash difference. I'm tempted to submit a patch that fixes this, but also its strange that the file is now different. this smells bad security-wise. <podiki>sleepylkw: first pull might take a while, what was the last message? but shouldn't be hours, can always kill it and try again <adanska>sleepylkw: if you check top, can you see if there is cpu utilisation? or has it hung? <sleepylkw>@podiki: last message is just the receiving objects 80% <sleepylkw>@adanska: seems to be hung, 0 cpu utilisation <adanska>that sucks. kill the process and try again :( <podiki>i think some of it will have been saved maybe? the receiving objects i'm guessing is the various guix files (a git checkout essentially) <iyzsong>adanska: it seems boost now redirect the download url to a html page of jfrog, which guix (or wget) can't handle. It doesn't seem like a security issue though. <adanska>iyzsong: okay, cool. I'll send a patch to try and fix this then, unless anyone else wants to <meaty>podiki: that fixes the version in guix, thank you! but as you said that version is seven years old anyways <meaty>should I still submit a fix patch? <podiki>Just the sdl2 change is all it needs? <meaty>podiki: yeah, that guix shell command you sent fixes it. it's still 7 years old, but at least it runs <meaty>the latest release, which I'm currently packaging, has a cryptic segfault I mentioned earlier <meaty>I even tried a few releases back, same issue <podiki>meaty: i'm happy to push the sdl2 change for the current version, but do remind me tomorrow if you don't see it (can't do it right now, deep in some guix home configuration changes) <jakef>perhaps the build is currently broken? i'm not sure how to check <adanska>iyzsong: submitted the patch as 75887! <xelxebar>For a good while, I frequently see network errors on long guix pulls or updates. <xelxebar>TLS error in 'write_to_session_record_port' procedure: Error in the push function. <xelxebar>Usually gets past that by re-executing the command, but it's annoying, since it means unsupervised upgrades are unreliable. <cdb-bugman>is "locale-definitions" in my operating-system declaration the way to add support for other languages and their symbols? im a bit confused on it <xelxebar>cdb-bugman: It's just the set of locales (i.e. for LANG and LC_*) you want your system to support. <xelxebar>That's not directly related to displaying symbols, which is more of a font thing. <xelxebar>A locale setting will tell applications what language, date format, collation rules, etc. to use. <xelxebar>And your installed fonts (and fontconfig settings) will make available the glyphs (i.e. symbols) for your languages. <rekado>jakef: the problem there was that the build node ran out of space <rekado>jakef: it would need to be restarted by someone, such as civodul. <jakef>okay, i'm trying to build it locally now in the meantime <rekado>I built python-tensorflow locally before pushing the update, but it takes a lot of space and more RAM than I have (so I had to restrict it to 2 cores) <jakef>roughly how long did it take with 2 cores? <rekado>but I also built jaxlib and jax from source <rekado>xelxebar: I see that too, but I don't know why this happens. I first assumed it was a problem with the firewall at the MDC, but it also happens when connecting to bayfront, which is hosted elsewhere. <jlicht>Alternatively, a RTFM pointer to how I could {be onboarded,onboard myself} would also be very helpful :-) <rekado>jlicht: do we have a jobset for this sort of thing or do you want me to create one? <jlicht>I _think_ I'm asking you to create one <rekado>surely this is covered by an existing team <xelxebar>sleepylkw: Looks slightly different. I'm not seeing the "died unexpectedly" but just a message that the build failed. <rekado>then I'd create a jobset for that team <jlicht>I was planning to kick off the node team at Guix days, so not yet at least <xelxebar>rekado: Hrm. So it's a mystery. Okay. If I get around to instrumenting some logs around this, I'll share. <rekado>jlicht: I suggest pushing this to a branch, then open a request for merge ticket (so that bayfront can also get busy). Once the branch exists I'd create a jobset on ci.guix.gnu.org. <jlicht>The request for merge ticket is just doing what I've seen zipping by on the ML before? <rekado>> To help coordinate the merging of branches, you must create a new <rekado>guix-patches issue each time you create a branch (*note The Issue <rekado>Tracker::). The title of the issue requesting to merge a branch should <rekado>> Request for merging "NAME" branch <rekado>the subject should match exactly, and the branch should exist. <jlicht>rekado: Consider me RTFM'd. Thanks! <rekado>the RFC process should make it easier to get up to speed on changes to the contribution workflow <rekado>I had missed these changes myself and made people unhappy by jumping the queue with my big R upgrades. <jlicht>Work + Real Life in a particular adorable and small form has made it hard to keep up. Thanks for being patient <rekado>you can view the queue of merges at qa.guix.gnu.org alongside the issues discussing the branches. <rekado>oh, yeah, portable humans do have that effect :) <sleepylkw>does anyone know how/if you can pass configure flags to emacs? <futurile>sleepylkw: you mean you're compiling emacs with `guix build`? <sleepylkw>futurile: ah thank you i think i see it now i assume I want a package variant? <futurile>sleepylkw: you can generally start by doing it on the command line, or in a manifest. Then beyond that you can create a package that 'inherits' <efraim>oh man, that was a big golang update <efraim>I think I dropped 53 golang packages from my channel, so now I'm finally under 300. I should see how many of them I actually "need" <efraim>not that I've worked at packaging any of the golang packages I was looking at for probably years <civodul>which made me think i should have stayed in bed <cbaines>civodul, do you have any thoughts on restarting bayfront? <cbaines>I can't seem to change the QA frontpage when reconfiguring, so I'm guessing the system isn't in a great state <civodul>cbaines: to “change” the QA frontpage? <civodul>i personally won’t be able to go on site before FOSDEM; there are two other people who can do it <cbaines>civodul, when I restart the service after reconfiguring, it doesn't change to match the configuration used <cbaines>the system config says revsion 29, but it's stuck on revision 27 <efraim>oh nice, guix style fixed "Package assert æ\x98¯å¯¹ testing å\x8c\x85ç\x9a\x84ä¸\x80äº\x9bç®\x80å\x8d\x95å\x8c\x85è£\x85" to "Package assert æ ¯å¯¹ testing å ç ä¸ äº ç® å å è£ " <meaty>Can anyone help me figure out why the pcsx2 package I'm working on keeps trying to build and test python(?) <futurile>meaty: you mean why it's building it rather than using a substitute? <futurile>meaty: cos, the reason it wants it is the 'python' is in native-inputs - or have I not had enough coffee yet? <meaty>futurile: I think it might be doing that, or it's somehow doing that as part of its build process <meaty>I put that in native-inputs to try and stop it from doing that <cbaines>meaty, use --dry-run and check if it says python will be built or not <futurile>and remove it from native-inputs - because it will definitely try and use python like that <meaty>cbaines: --dry-run only says that curl will be built and tested <cbaines>that suggests that you don't have a issue with the python package at least <cbaines>maybe share the start of the build log if you're still having issues <meaty>cbaines: right now I'm waiting for curl to finish its test suite--at least it seems that python build wasn't vendored, cos it's not doing it again. So hopefully once the prereqs are built rebuilding will be fast enough to get the final product working and then pare down the process <meaty>okay it said only curl, but now it's building and testing CMake and it also built zlib and ninja? <civodul>futurile: yay! LGTM, so do ping me or other committers when you want to publish it <civodul>cbaines: did ‘reconfigure’ complete? did it say that it loaded replacements for the service? <cbaines>civodul, it does complete, but I do get the "no code for module (shepherd service timer)" error <civodul>in that case, we can use ‘herd eval’ to load the new qa-frontpage service anyway <civodul>and after FOSDEM we can reboot without fear <meaty>Could changing the C/++ compiler via #:configure-flags have triggered a rebuild of the whole dep tree with clang/++? <civodul>so now if you run ‘herd restart qa-frontpage’, you’ll get the new one <cbaines>civodul, nice, thanks, that's worked :) <meaty>yes, I'm 90% convinced that this package def tries to bootstrap almost an entire system. it's now built poppler, cups, qpdf, and whatever else... idk what's going on <cbaines>meaty, maybe use --no-grafts along with --dry-run and check what that says <meaty>cbaines: It gives a humongous list of packages <meaty>like, all categories, ghc, ruby, rust, texlive, everything <cbaines>are you working in a guix.git checkout or not? <meaty>cbaines: I'm working in a checkout <cbaines>you need to work out if this is because of a change you've made, or whether there are just substitutes missing <cbaines>meaty, check git and see if you've made any changes to packages other than the one you're working on <meaty>cbaines: The only changes from master are the added pcsx2 definition and updated libpng... oh <cbaines>Building the following 11519 packages would ensure 26719 dependent packages are rebuilt ... <cbaines>so I think what you're seeing is expected <meaty>do you think the right way to go about it is to create a seperate def for the libpng pcsx2 needs until I'm done hacking on its definition? <cbaines>meaty, assuming that works, it sounds like an OK approach <jlicht>rekado: pushed the branch, sent the email; could I bother you into checking if things went well, please? <sleepylkw>has anyone figured out how to install emacs with native compilation? I found flatwhatsons package but it looks to be tracking 28 <Rutherther>sleepylkw: what do you mean by that? emacs in guix is with native compilation. Do you mean how to get guix emacs packages native compiled? <sleepylkw>Rutherther: oh maybe im confused... In the installation output I see " Does Emacs have native lisp compiler? no <Rutherther>sleepylkw: unless you are using emacs-minimal, native compilation is enabled <Rutherther>or of course if you just overriden the original arguments instead of appending to them <apteryx>neat, our biggest annoyance (?) with GNU Debbugs is a known bug: bug#15361 <apteryx>civodul: Hi! I haven't had much luck with 'reboot --kexec'; my screen turns black and that's it. Does it work on your machine? <rekado>jlicht: thanks, I'll create a jobset for node-team <jlicht>apteryx: I need to enter my luks password once, followed by enter, before I see stuff again <jlicht>apteryx: so I run reboot --kexec, wait for the screen to turn black, give it a good 10 seconds, and then input password followed by enter, and then I'm looking at a tty with login: <civodul>apteryx: actually, it’s the same on my laptop! i’ve used it successfully on servers though <civodul>jlicht: oh so LUKS is waiting for the password while the screen is blank? <dthompson>rekado: ooh is that the node update that's building? yay! <sneek>Welcome back dthompson, you have 1 message! <sneek>dthompson, jlicht says: the support for publishers in haunt is _really_ nifty! <squid64>Just got GNU Guix System installed 2 days ago and I think I found my new favourite distro, integrates very well with Emacs too somehow. Reminds me of NixOS but it's also different. Almost all the FSF approved distros were distros I didn't really like except for maybe Parabola but this one I actually like. <squid64>It's also good that it has an easy installer. I like the more advanced distros but I like to dead with that kind of things after install rather than in the install. <apteryx>civodul: I'm guessing some GPU related problem then :-/ <apteryx>good to know that my system is not a snow flake <squid64>Is it before grub shows you are getting that? <apteryx>jlicht: oh, just saw what you wrote about the LUKS passphrase; that could be it, I do use full disk encryption. <squid64>I also sometimes just see a black screen at boot but I just enter encryption password and it works. <apteryx>squid64: I typically see a prompt (ugly, but functional) at boot, but not when using the recently added 'reboot --kexec' (the screen is pure black then) <squid64>Yeah I tried reboot --kexec once and I had a black screen like monitor turned off and all. I just never reused it after that happened. <squid64>I didn't try entering encryption password or anything in that state though as I thought my PC just froze or something. <civodul>looks like that kexec thing is not entirely satisfactory :-) <jlicht>dthompson: no clue why sneek bothered you here instead of over at #guile though... <dthompson>sneek doesn't track which channel the message was for unfortunately <Deltafire>i had the black screen with --kexec, but i tried it a few days ago and it actually worked <Deltafire>although only saves a few seconds, so why bother.. <jlicht>for me it still saves a lot of time, as doing the grub LUKS thing takes extremely long on my machine <Deltafire>ah yeah, guess it makes sense in that situation <apteryx>I've submitted this refinement to the maintainer of Emacs Debbugs, but in the meantime, if you are bothered that replying to a patch for reviewing looses the patch number, customize your debbugs-gnu-summary-keep-subject variable to "\\(?:\\[\\(?:[^[:space:]]+[[:space:]]\\)?PATCH \\(?:[^[:space:]]+[[:space:]]\\)?\\(?:v[[:digit:]]+[[:space:]]\\)?[[:digit:]]+/[[:digit:]]+]\\)" <apteryx>lilyp: I meant to review to your large series, but I see it's merged already! One less thing to do ;-) <apteryx>I did look quickly through it, and it looked like mostly version bumps, so no big deal. <FrenchNewbie>hello, i found some time today to try my first derivation of my guix system install, i tried installing bat and btop throught the config.scm by adding the packages name but i'm having an error "wrong type to apply: #<package bat@0.24.0 gnuj/packages/rust-apps.scm:275 7fe14c02e370>" can anyone help me debug that ? <FrenchNewbie>i had no error when installing them throught "guix install" <bjc>i thought i saw a post on the mailing list recently about running guix with tmpfs root, but i can't find it in the archives. anyone remember this? <oriansj>FrenchNewbie: you need admin and rust-apps in your use-package-modules section of your configuration <oriansj>if you did guix package -s btop ; you'll see the location which you need to use <oriansj>also %default-log-expiry appears not to be defined anywhere (as far as grep -iR is concerned) <janneke>oriansj: hasn't log rotation moved to the shepherd? <oriansj>janneke: looks like it (but I do explicit guix configuration and the lack of documentation is not helpful) <anemofilia>civodul: Hi, I noticed you have upstreamed guix services for shepherd-timer and shepherd-transient. I was thinking that we maybe could rename shepherd-timer to shepherd-transient-timer or something and then make a shepherd-timer service to allow easy declaration of persistent timers a la mcron, WDYT? <sobol>hello all. has anyone been able to install guix on a lenovo thinkcentre 720s? i tried but i just get an error about unknown device with an id string that doesn't help me. thanks. <ajarara>is it UUID? on boot? My guess is copied and pasted config without updating the disk ids <ajarara>sobol: if it is on boot, reboot back into the guix live CD, enter a shell, run lsblk -o NAME,UUID and compare with what was installed. <sobol>thanks @ajarara. ill give that a try. <ajarara>sobol: no prob, lemme correct myself a bit: guix install CD has a shell, you just have to do like Ctrl+Alt+F1-7 (on an install F8 is where I boot to). If you went through the install CD I _think_ the config that you already installed is in /etc/config.scm <sobol>oh i don't have guix installed at all on the machine. i bought it off ebay without an OS. i have installed nixos just to make sure that everything works as expected. when i boot my usb drive and select the "install guix" option, that is when I get an error and dropped into guile <sobol>i am completely new to guix. i have only installed it in a vm before but now want a dedicated desktop with it :) <ajarara>oh. I'm definitely not sure what could cause that pre-install. Could you give the exact error you see in a paste? <sobol>yea, i think i DL'd the latest iso. i am now DL'ing the standard and will give that a shot and get a paste of the error if it happens again. <ieure>sobol, Haven't run it on that specific machine, but have put it on an older M93p, and a newer ThinkStation P340, both worked fine, other than the pain of no WiFi drivers. <ieure>sobol, I've found that the simplest way to get Guix installed is to turn on wired tethering on my phone and plug it into a USB port, it shows up as a USB Ethernet device. Can turn off the cell connection to make sure you're using WiFi and not eating your mobile data. This is on Android, no idea if iPhones can do this. <sobol>@ieure thanks for that info. the machine is wired into my switch so no wifi issues for this install. ill give it another shot with a fresh iso dd and report back. thanks. <ieure>sobol, Are you aware of Ventoy? <ieure>Ventoy is good stuff, you install it onto a USB stick, then can mount it as a FAT filesystem and copy images onto the filesystem. When you boot off it, it gives you a menu that lets you pick one of those images. <ieure>Means you don't have to dedicate one flash drive per thing you want to boot, or constantly overwrite one with different images. And no messing around with dd or root permissions, just mount, copy, unmount, boot. <sobol>i wasn't. ill def look into that. great tip. thanks! <dariqq>oluminol: i think this means that there is an error in GUIXito module but the error is not really helpful <dariqq>(the problem is the same as #75882 where guix would advise to add (use-modules bla ) after failing to eval bla without reporting the error which creates a bad user experience) <meaty>are the "qt6 linguist tools" packaged anywhere in guix? <coyotes4ys>i had openbox questions yesterday and we were talking. <coyotes4ys>so the guix build openbox is on my hard drive just inside the package manager? <ekaitz>coyotes4ys: the guix build command returns a path <ekaitz>the contents of the package are there