IRC channel logs
2024-02-21.log
back to list of logs
<Kingsy>podiki: you around? I have just found your dotfiles and something isnt working for me. was hoping you could spare a second, basically I want to jut exec dbus-run-session in my xinitrc but it says its not found. and I was wondering if there was a package or something I am missing <Guest96>Guix only allows free software. I had assumed that the qualifier here was "can be combined with GPL3 and conveyed", but Guix includes GPLv2, which cannot be combined with GPLv3 code <Guest96>I'm looking at Holochain and I'm wondering if it can even be included in a FSF approved OS <the_tubular>What's the difference between GPL2 and GPL3 ? I just know that Linus hates V3 for some reason <Guest96>The main problem with the holochain license vs GPL is clause 4.2 in the holochain license <peanuts>"Frequently Asked Questions about the GNU Licenses <Guest96>Maybe this is a question for the guix mailing list? I have connections in the holochain community, so if there is a license problem which would cause all fsf distros to reject holochain, this should be figured out now <Kingsy>anyone seen this error before -> Settings schema 'org.gtk.settings.filechooser' is not installed <Kingsy>it appears when I close the prferences dialog on xfce4-terminal. after which the terminal closes. <cow_2001>Can't exec "aclocal": No such file or directory at /gnu/store/kngd1h8b65kdnawd9mfxarwhkaxwq0lz-autoconf-2.69/share/autoconf/Autom4te/FileUtils.pm line 326. <snape>ieure: how is mullvad browser "not great for daily use"? <sneek>snape, janneke says: also, i checked systemd's closure size in nixos, it's bigger than guile: 147MB <PotentialUser-59>Any ideas why dhcp-client-service-type isn't sending the system host-name to the DHCP server? <podiki>Kingsy: dbus-run-session comes from the dbus package. in my case, probably from using %desktop-services in my system configuration, or from other dbus services i use <snape>janneke: thank you for this research, this is interesting <snape>Kingsy, Mullvad Browser is great for daily use <snape>(just correcting what's been said) <oriansj>is there any reason why all linux-libre versions build the exact same linux-libre-documentation version? <jamesvasile>Hi all, I have newly added guix to my debian testing machine, installed fbreader via guix, and running it gives me a seg fault. So I rerun and get the same result. I do thi s a few times and... now it works. I've run it and exited a few times now and it just seems to randomly seg fault. Where does one report in a guix package like that? Thanks. <iyzsong>jamesvasile: should report as a email to bug-guix@gnu.org <jamesvasile>In the mean time, I'll... start it a bunch of times until it lets me read :P <cow_2001>checking for libvigra_c... configure: error: libvigra_c not found <cow_2001>the modify-phases is from the existing guile-cv version 0.2.0 definition in guix <imadnyc>Hey, does anyone know why I'm getting this error? My home config: <imadnyc>Forgot IRC doesn't have markdown - using this through matrix bridge <mange>I think the function name is specification->package, using the singular "package", not "packages". <dodoyada>(sorry if I posted this already) I'm trying to make a contribution for the first time. When I get <dodoyada> to the step `guix pull --url=/path/to/your/checkout <dodoyada> --profile=/tmp/guix.master` in the patch guide I get "guix pull: <dodoyada> error: Git error: cannot locate remote-tracking branch <dodoyada> 'origin/keyring'". Did I miss a step? Also, what's the deal with <dodoyada> the /tmp/guix.master profile, should it be generated or is that a <mange>Have you got the keyring branch in your local checkout? Running "git fetch origin keyring:keyring" from "(guix) Building from Git" on your checkout might help? <mange>Assuming you're following the steps in "(guix) Submitting Patches" you might not have done this. <mange>The --profile=/tmp/guix.master is just there to run guix pull without clobbering your actual profile. You could use any path you want, and you can delete it once the pull is complete (or you could use the guix in it, if you want to test your changes after a pull). <nckx>oriansj: Can you clarify? Linux-libres shouldn't be building documentation at all, that's the point of the separate linux-libre-documentation package. And this package exists only for the latest kernel. <nckx>You could always try turning l-l-d into a procedure that takes a kernel. I stopped doing so precisely because the build's expensive & the differences too minor. <nckx>In better news, the build fix has made it to master. <bdju>thanks to whoever packaged aerc, I had been wanting it for a long time. I was using it over ssh on another machine until now even. <bdju>ah darn html emails don't show up right. I'd just copied over my config from the other machine <bdju>I get the following error: /gnu/store/q7h7ldxgm0vya6128g5sspmwd8wac61j-aerc-0.15.2/libexec/aerc/filters/html: line 5: exec: socksify: not found <nckx>bdju: Packaging bug aside, does manually installing dante get you past this error? <nckx>Yay. I'm patching aerc properly here. <nckx>…xcept notmuch won't build here. <nckx>It should reply to Bob but replies only to Charles. OK. <ulfvonbelow>do we have any ocaml wizards awake right now? I can't seem to figure out how to get dune to link to a system library without thinking it has to build it first. For example, what I'm trying to package (https://github.com/grievejia/pyre-ast) needs to link to a patched python. The way it wants to do this is by vendoring an entire copy of python and building it from scratch, even though the actual patch is tiny. I'm trying to unvendor it by <ulfvonbelow>deleting lib/taglessFinal/vendor and then modifying lib/taglessFinal/dune to use the environment to find headers and libraries for a separately-packaged patched python. But it seems to think that (foreign_archives python) means that it *needs* to build lib/taglessFinal/libpython.a. When I remove the (foreign_archives python) part it builds successfully, but during the check phase it yields 'undefined reference to <ulfvonbelow>`_PyParser_ASTFromString''. It seems to be attempting to compile lib/taglessFinal again, but this time without the -lpython flag that was specified in lib/taglessFinal/dune. I'm at a bit of a loss as to how to make this work; the build system seems to only support "build and link this C code", not just "link this C code", which seems like an absurd oversight, so I'm sure I'm missing something <peanuts>"GitHub - grievejia/pyre-ast: pyre-ast is an OCaml library to parse Python files. The library features its full-fidelity to the official Python spec, as well as its adoption of tagless-final style." https://github.com/grievejia/pyre-ast <ayatsfer> (or (git-predicate (dirname (dirname (current-source-directory)))) <ayatsfer>how could I extend it to also exclude files if they end with .scm? <ulfvonbelow>ayatsfer: #:select? (lambda (f) (and (vcs-file? f) (not (string-suffix? ".scm" f)))) <fnat>Guix-wise, I'd imagine there's not much to do with a Raspberry Pi 3B+ (1GB ram)? Or is there a way to flash an image and bring some Guix-magic to this little RPi that I've lying around? <adanska>perhaps something using `guix system image`? i think on the forbidden channel theres raspberry pi drivers you can put into a config to get it to work <adanska>fnat, i know its been done and continues to be done since ive heard of people using guix rpis and also we have armhf substitutes (not sure how healthy they are though.. depending on how they are you might have to set up your parent install to have cross-comp support for armhf so it can build the packages into your image <adanska>this is all hypothetical, ive never tried it myself. <fnat>That sounds promising, I hope this applies to my poor old 3B+ model. :) <ulfvonbelow>the 'undefine reference' error only seems to appear when passing --release to dune <oriansj>nckx: thank you for clarifying about the linux documentation. it makes sense. <ulfvonbelow>actually it might just be that --release disables deprecation warnings, allowing the build to proceed farther, at which point it gets the 'undefined reference' error <snape>janneke: I just checked, guile-static-stripped@3.0.9 is 54.1MiB ... <janneke>snape: yes, 2.0.9 is about 14MB; not sure where that huge difference comes from <marcc_>Heya, I tried to send an email to help-guix yesterday but I notice it isn't showing up in the archive. Do I need to be subscribed in order to send emails there? <snape>marcc_, you don't need to be subscribed, but if it's your first time there's human validation, so you just need to wait a bie <iK0u>nckx Good news. I managed to reproduce the same derivation as you by doing it inside a guix vm I generated on the host that was giving the wrong one. The one you produced is also the same as one of my devices. I believe the problem was that I am using NixOS on the device that gave the wrong derivation and, despite the user being on the correct commit the root user's shell environment was not configured properly for guix, I didn't think of it because I never used <iK0u>the root user with guix before. <iK0u>This is the correct derivation: /gnu/store/sm650wj4biafx9d148bzmik728qh5yqq-image.iso.drv <iK0u>However, neither of us managed to get the same derivation as the guix ci <iK0u>Does anyone know if it is possible to figure out exactly what command/flags/method a given build on cuiraiss is using so we can produce the same derivation for ie - the installation iso <snape>wow being able to download the output like this is nie <civodul>how’s Debbugs today? my patch doesn’t seem to make it there <iK0u>snape The download button for it doesn't actually work but.. it really is downloadable if you modify the link a bunch :P <old>I try to use the vcs-file? predicate from guix cookbook. However, current-source-directory always returns #f. Any idea? <snihil>hi Guix, I've attempted to package wayfire and it builds fine now. The problem I have is that it fails at the validate-documentation-location phase and I don't get why. Here is the log, in case anyone wants to take a look: https://x0.at/Ih_1.log <snihil>I have also attempted a gc of all wayfire-related store items and building again, just in case. <civodul>snape: uh, maybe some congestion at lists.gnu.org? we’ll have to be patient <civodul>plus Sharlatan’s astronomy patch series <PotentialUser-59>My shepherd service died after a reconfigure(guix system: warning: while talking to shepherd: Connection refused), and now I can't even reboot my system. If I start shepherd myself and run reboot, it just kills the shepherd process. What can I do short of physically shutting the system down? <PotentialUser-59>(reboot and halt both give error: /var/run/shepherd/socket: Connection refused) <efraim>is there a way to refer to the lib output of gcc-final from inside a package definition? without adding gcc:lib as an input? <dariqq>should it be possible to subscribe to guix-devel? tried a couple of weeks ago (both via mail and the webui and again today) and i never get the mail to confirm my subscription. <ekaitz>dariqq: it should be possible, i am subscribed <futurile>man I do not understand Debbugs, and specifically GNU's - absolutely no commands work from the documentation - and the commands that do work shouldn't <dariqq>or maybe my subscription request would need to be approved? but that would be odd as it worked for the help ml without issue <ekaitz>futurile: I had a hard time understanding i needed to \nquit <futurile>and consequently the bugtracker is a dumpster fire of out of date stuff <ekaitz>futurile: and i'm pretty sure i'll do it wrong again the next time i want to use it <futurile>ekaitz: do you happen to know how to either (a) archive or (b) close a bug? I tried control@ and nnn-maint@ and can't get either to work <ekaitz>futurile: i think it was something like close\nquit (being a \n an actual newline) <ekaitz>futurile: choose an issue and let's spam it with our tests lol <ekaitz>i searched for `quit` just to check the syntax hehe <ekaitz>it's `close number` newline `quit` <ekaitz>i want to clean my old issues too... i'll should spend some time doing that <podiki>also mailing nnn-done@ will close a bug <ekaitz>podiki: true, that's what the manual says: When a bug is resolved, please close the thread by sending an email to ISSUE_NUMBER-done@debbugs.gnu.org. <podiki>that's what i usually do, in a wide reply, as then there is a message of why it was closed; but i guess if there was such a message and just forgot to add -done, using control to clean up a bunch is nice <podiki>on this note, we have an epidemic of missing messages: people (myself included) will send to nnn@debbugs which goes to exactly no one <podiki>in replies to a thread. if it is just the submitter doing things there's no one else to send to of course <iK0u>can "guix archive" be combined with guix system image? <podiki>i know better now but still see messages all the time that were not received because of this. rather unintuitive i would say, i think most expect nnn@debbugs to be a mailing list essentially <dariqq>mailman is weird. Sent another subscription request per mail and it worked instantly now. Never got an answer via the webui. Not sure what i did wrong a month ago. <podiki>while i'm here, will reach out on guix-devel but any pending patches/wishes for mesa-updates? <podiki>my plan: update libdrm, cairo, mesa, vulkan (probably); see about enabling the nvk driver in mesa but that requires rust so just x86_64? <futurile>anyone know what's happening with core-updates? I've lost track of how 'soon' it is to going in <podiki>there's some discussion on guix-devel; glibc update seems mostly good, some updates needed around a crypto library in guile if i remember, that needs to be resolved how to handle dependencies <podiki>there's no time frame, just needs to be in a workable state, i'm sure help is appreciated :) <futurile>I'm *waaay* too busy being messed around by an email bot that has a mean sense of humour to actually do any helping! <efraim>podiki: one option is supported-package? to gate the package to supported systems <podiki>yeah, that's what we'll have to do. but we'd expect a rust-based driver in mesa to only build for us on x86_64? or we now have some aarch rust too? <efraim>we currently have rust working on aarch64 and after the rust-team merge it'll be working again on riscv64 <podiki>someone via another channel has a supposedly working mesa with nvk, so i'll try to get them to submit a patch here <efraim>we're close on powerpc64le, currently it fails to build 1.73 <podiki>yeah, will be a conditional input on rust (and whatever else it needs) and the configure flag <podiki>sounds like it is a decent step forward from plain nouveau? (thankfully no nvidia hardware on my guix machines) <efraim>my nouveau driver keeps on crashing on me, i wish I knew how to reload the driver or something <podiki>sounds like you will be a tester! :) <podiki>i take it that is a vote for enabling it as an option (i don't think it replaces nouveau? not sure how it works) <efraim>I even bought a new old GT730 and it was still crashing <civodul>cbaines: 👍 on the store and database patches! will look into them in the coming days <podiki>anything i should be aware of if i try to update cairo? seemed a few things in release notes (moved to meson to build, enabled stuff we already enabled, dropped xml backend) <podiki>(i will send a patch first but will build locally to test of course) <efraim>I would check with the gnome team about cairo <podiki>right, they are doing big updates maybe they would want it there actually <futurile>civodul: do you know if you have to on some secret list to send emails to control@debbugs.gnu.org. I'm sending the same commands that you sent to close bugs and it just ignores me. (Yes I can close by sending to nnn-done, but for old bugs I was hoping to not spam people) <podiki>i don't think so. but there has been mailing list wonkiness recently <efraim>IIRC the FSF mailing lists everything were down for ~68 hours last weekend <futurile>OK, I'm going to switch to doing something else, not feeling productive :-) <efraim>if I'm reading this right the ld command for linking mold on powerpc-linux has 2.7GB of virtual memory and is using 847MB so far <efraim>will have to look into what to do with that <civodul>futurile: i don’t have magic powers no :-) but yes, the list is super slow it seems <civodul>i actually send control messages from debbugs.el <civodul>which is nice because otherwise i can never remember what to send and where <reedm>Not sure if this is a better question for the guile or guix channel, but I am trying to clean up my substitutes code in my config.scm, and I'm having a hard time quoting the public key. Does anyone know why this is invalid scheme: '(ecc (curve Ed25519) (q #ASDF#))? <reedm>The googlability of this error is tough, since the "#" symbol has so many colloquial names, and is usually ignored if typed in directly <Gooberpatrol66>if a package is not listed on crates.io, is it still possible to import its dependencies somehow by passing its cargo.lock file to guix import? <iK0u>'(@ (gnu packages guile) guile-2.0)' evaluates this one packages <iK0u>how would you evaluate everything in gnu packages instead? <iK0u>For example if you wanted to guix build -e everypackage <iK0u>I tried doing guix build -d '(@ (channels) guix)' and various iterations, can't seem to figure it out, I don't really know much about scheme yet <iK0u>the cuirass service is able to build everything in a channel with " (build '(channels my-channel))" according to the manual <iK0u>but this syntax did not work with guix build -e '(channels guix)' <nutcase>Hello guix, which packages does my shell need, when I'd like to have PkgConfig to build some C software, namely https://github.com/Ferdi265/wl-mirror/? While running cmake -B build in a `guix shell make cmake gcc-toolchain`, I get the error "Could NOT find PkgConfig (missing: PKG_CONFIG_EXECUTABLE)" <apteryx>iK0u: I know 'guix refresh' can take a module as input and refreshes every public package in it <apteryx>you could check the code that implements that underneath <nutcase>apteryx: thanks! That got me further (shame on me, that was too obvious). Next issue: wayland-client is missing: "No package 'wayland-client' found". Including rust-wayland-client in the shell, didn't help <janneke>apteryx: yay, your dependency fixes also fixed the hang I got stuck on building a (bare-bones) system <nutcase>apteryx: nevermind, I managed to build wl-mirror <nutcase>is it worth to create a package and submit this? <Guest79>getting into the installation of packages at the moment <mekeor>Guest79: nice, welcome! how can we assist you? <Guest79>wonder how easy/difficult it would be to contrib packages I miss at the moment <mekeor>Guest79: writing packages can take 5min or 5months. i'm not sure about this one. <Guest79>That's what I expected :-) guess it depends also on phrasing the proper question(s) <cobra>speaking of canoeboot, has it been updated for the latest libreboot release? <Guest79>it has some dependencies on ruby, sqlite, nodejs which are already packed in guix <juanpablo>Hi, I recently installed guix with the system crafters image. I enabled 2 desktop environments (gnome and exwm). For now whenever I log in I get the gnome desktop, how can I switch to using the exwm one? <Guest79>I'm going to give this a try. Would it be ok if I (occasionally) pastebin some of my efforts for review/help? <nutcase>In a package definition, how can I refer to the directory of an input package? I have '(inputs (list ... wayland-protocols))' and need to provide '(arguments (list #:configure-flags '(list "-DWL_PROTOCOL_DIR=<directory-in-store-of>-wayland-protocols-<version>")))'. <nutcase>I guess, it's something with (append) and (guix build). But probably there is a better way? <apteryx>janneke: dependency fix? which commit was that? :-) <ieure>nutcase, the name of the package is bound to the store location in the build gexp, so (string-append #$input-package-name "path/within/that") <ieure>There are a bajillion examples of this in the package definitions shipped with Guix. <janneke>apteryx: dunno, any of them that i found this morning around %final-inputs? <janneke>ACTION can't imagine it was the merging of master that fixed it... <nutcase>ieure: (string-append "-DWL_PROTOCOL_DIR=" #$wayland-protocols "/") gives me a "Unbound variable: ungexp". wayland-protocols is in my (inputs (list ...)) <apteryx>rekado: for hydra-guix-129, I think to offer it as offload machine, users would need to be able to ssh in, and we'd also need to authorize their machine host key. That would make the machine unsuitable as a substitutes source/build machine. <nutcase>ieure: also not successful: '(list #:configure-flags '(list "-DFORCE_SYSTEM_WL_PROTOCOLS=ON" (string-append "-DWL_PROTOCOL_DIR=" (assoc-ref inputs "wayland-protocols"))))' <juanpablo>[help] How do I change from one desktop environment to another in my config? <iK0u>juanpablo do you have a login manager? like gdm or sddm? you can select which environment from there <iK0u>there is a button to select the session <ieure>nutcase, Your build steps aren't in a gexp. You can look at pretty much any C/C++ program Guix has a package definition for to see how to set this up. <ieure>I'm on my work computer, which doesn't run Guix, so I can't find one to link you to, but there are gobs. <juanpablo>iK0u: not sure I have gdm, I'll check now. If not how do I add it? <iK0u>juanpablo when you login, you do so from a graphical display right? <iK0u>if you have any of them you can select easily with a button <iK0u>just select it before logging in <juanpablo>IK0u yes I get into an ubuntu like login but I can only select the user, not the environment <iK0u>do you see a gear/cog icon? <iK0u>yw, I didnt know where exactly because I use sddm instead <Kingsy>if I want to run xorg-server via xinit on guix. what extra packages do I need to run gtk apps and others? do I need to install gtk? dconf? I have elogind and %desktop-services% <ieure>Kingsy, I don't think you need to install anything other than the packages for the programs you want to run, all the inputs get brought along with them. <ieure>Kingsy, Hm, not sure then. I haven't had this problem, but I guess I have Gnome installed, even though I don't use it regularly. <ieure>But I can log into my EXWM desktop and run gtk stuff without issue, and haven't done anything special for that to work. <Kingsy>going to uninstall a bunch and start again. see what state it is in <Kingsy>ieure: no I could run the apps. but with xfce4-terminal for example if I right clicked -> preferences then clicked close on the preference window the app crashed. <Kingsy>so it was kinda subtle the way it was broken <Kingsy>ieure: when using desktop-services how do I remove gdm? so I don't have a login manager? just curious? thi is my current config -> https://bpa.st/TXLA <- so you see I have tried a modify-services on there. but after a reconfigure and reboot and I still got gdm <hatim>When running emacs, I am getting a Gtk warning: cannot open display: :0 <hatim>I have these env variables setup, <hatim>How do I start debugging this issue <dariqq>Kingsy: the set-xorg-configuration will also add gdm by default. <Kingsy>dariqq: wow... its like a plague <podiki>Kingsy: i think you saw my dotfiles so that is an example of not using gdm (i've used sddm and lightdm) <podiki>hatim: are you using emacs-pgtk? btw, the default DISPLAY i've seen on guix has been :1 <Kingsy>podiki: well. I got xinit working, so I really just wanted to not bother with any login manager. <podiki>i used to just startx myself, but not on guix <dariqq>anyone successfully cross compiling anything? When i try for i686 it starts building a lot of the bootstrap gcc stuff and then (eventually) fails at glibc-mesboot during the setenv phase <Kingsy>podiki: what I mean is. its not possible to delee gdm without adding another login manger to rplace it? <Kingsy>podiki: its not possible it seems. <Kingsy>so its either base services, or a login manager. <hatim>podiki I wasn't, emacs-pgtk works... is it because I have xwayland, that I have to use PGTK? <podiki>hatim: I think using pgtk means emacs runs on wayland, not xwayland <podiki>though plain emacs should work as well, just via xwayland (e.g. text scaling a la X rather than Wayland), perhaps you have some env issue then <hatim>podiki ok, was just about to ask if there was a way to make it run on xwayland <podiki>xwayland, as i understand it, is for applications which don't run on wayland directly <podiki>so if you want emacs on xwayland, don't use the pgtk version. perhaps check that DISPLAY is set correctly. as i said, mine is set to :1 and I don't touch it manually anywhere <podiki>(xwayland = x server in wayland roughly? i'm new to wayland) 