IRC channel logs

2020-08-03.log

back to list of logs

<roptat>if i could override IBUS_LIBEXECDIR, I think I'd be able to set it to my profile's libexec which contains both ibus and ibus-anthy stuff
<roptat>yes, it worked!
<roptat>but I still can't type japanese T.T
<roptat>maybe a reboot will do
<bdju>how can I find which revision(?) of guix a package was added in?
<leoprikler>git log
<leoprikler>searching for "<package>: [nN]ew" should usually bring up the commit
<bdju>I guess I have to have a local copy of the guix repo then. do I just git clone 'https://git.savannah.gnu.org/cgit/guix.git/'?
<bdju>oh wait I see a guix dir in my downloads, maybe I do have one
<pkill9>you can search the cgit instance as well
<lfam>Is there a way to build 32 and 64-bit ARM vm-images from x86-64?
<roptat>ok, ibus-anthy is missing a dconf schema, which is present in the git, but not in the release tarball
<roptat>but it's present in the newest version, let's see if updating the package will do the trick...
<roptat>which package contains getopt?
<alextee[m]>roptat: are you fixing the japanese input? :O
<roptat>yes
<alextee[m]>it's one of my long standing problems with guix
<alextee[m]>sending much positive energy and vibes your way!!
*lfam sends linux-libre 5.7 update patch
<roptat>ok, the newer ibus-anthy can register preferences, but it still doesn't solve the issue
<leoprikler>I may have missed a few things, what exactly is the issue?
<roptat>basically I can't type japanese
<roptat>I can open ibus-setup and add anthy there, then I can switch between French and Anthy input methods with the keyboard shortcuts, but it still inputs with my French layout
<leoprikler>is this GNOME or elsewhere?
<roptat>openbox
<leoprikler>hmm, how does openbox handle input methods?
<roptat>I see deadkeys are underlined like it's supposed to be in Japanese input mode, but the rest just inputs directly, so I thought it might be a configuration issue in anthy
<leoprikler>deadkeys are underlined in ibus generally
<roptat>but our ibus-anthy can't save preferences in dconf it seems, so I just fixed that
<leoprikler>like even with my German keyboard ` gets underlined
<leoprikler>Also C-S-u
<leoprikler>ahh, did you set anthy to kanji mode?
<leoprikler>IIRC it does not start out that way
<roptat>how do I set it to kanji mode?
<roptat>arg, anthy disappeared from ibus-setup
<roptat>why is IBUS_COMPONENT_PATH not provided by guix package --search-paths anymore?
<roptat>oh because I installed it from the store path, I'm stupid
<leoprikler>not sure, in GNOME there is a switch
<leoprikler>(with kanji mode I meant one of hiragana/katakana)
<roptat>mh... I switched back to the generation I was before trying to get japanese and anthy is still gone...
<alextee[m]>it doesnt work in most apps in gnome btw
<alextee[m]>but weirdly, japanese input works when you press the windows key to search for apps
<leoprikler>I do have a working GNOME configuration fwiw
<alextee[m]>that's one of the few places it works
<alextee[m]>i sometimes just type stuff there then cut it and exit and paste it where i want
<alextee[m]>that's how to type japanese in guix
<alextee[m]>:P
<leoprikler>It even works in Epiphany, where display is still broken ;)
<leoprikler>ibus/gnome is really not difficult
<alextee[m]>it used to work in that qt matrix client
<alextee[m]>quaternion
<alextee[m]>seems to work at specific places only
<alextee[m]>other places it enters english characters instead
<leoprikler>I just extend the gnome metapackage to propagate ibus and ibus-anthy
<alextee[m]>that makes it work everywhere?
<leoprikler>then add those two lines to .zprofile (or .profile, whichever you source):
<leoprikler>export GUIX_GTK2_IM_MODULE_FILE="/run/current-system/profile/lib/gtk-2.0/2.10.0/immodules-gtk2.cache"
<leoprikler>export GUIX_GTK3_IM_MODULE_FILE="/run/current-system/profile/lib/gtk-3.0/3.0.0/immodules-gtk3.cache"
<leoprikler>and it's instant すごい
<leoprikler>not sure if "everywhere", but in all of my GTK+ apps, plus that random Qt one it does
<leoprikler>If it also works with Openbox we might want to extract this into an ibus-service-type
<roptat>help :(
<roptat>I even tried wiping dconf settings
<roptat>nothing anthy-related in .cache or .local
<leoprikler>so just to recall, it completely vanished from ibus-setup, right?
<roptat>yes
<roptat>I switched back to a generation where it was present
<leoprikler>what about your environment variables?
<roptat>when I try to add an input method, Japanese only lists ibus' japanese, but not Anthy anymore
<roptat>IBUS_COMPONENT_PATH looks correct
<roptat>GUIX_GTK{2,3}_IM_MODULE_FILE are also set correctly, but I don't think they influence that part of ibus
<roptat>also, everything should be the same, as I use the home manager for my .profile (so not modified), and I switched my default profile back
<leoprikler>sounds pretty harsh
<leoprikler>IIRC you once got it to work by setting the component path to the store directly, right?
<roptat>hm... I don't think so
<roptat>strace says ibus-daemon looks in ibus' store path directly, ignoring IBUS_COMPONENT_PATH
***flor_ is now known as flor
<roptat>now that I think of it, anthy was still there for some time after IBUS_COMPONENT_PATH disappeared, so there must be some kind of cache somewhere
<roptat>and I can't find it
<ryanprior>Potentially a big win for Guix: Rome is a vanilla javascript with no third-party library dependencies. It's a linter, compiler, bundler, and more for JavaScript, TypeScript, HTML, Markdown, and CSS.
<ryanprior>A rome-build-system could potentially allow building many nodejs packages which normally would depend on a massive snarl of build-time dependencies.
<ryanprior> https://github.com/romefrontend/rome
<ryanprior>I'm gonna work on packaging Rome and creating a rome-build-system. Anybody who's interested in following along feel free to ping me :)
<leoprikler>I'll expect you to be done within a day 😉️
<raghavgururajan>Hello Guix!
<apteryx>hello raghavgururajan
<raghavgururajan>apteryx: I accidentally `rm ~/.config/guix/current` instead of `rm ~/.config/guix/gcroots/current`. What should I do to restore/recover?
<Noclip>raghavgururajan: You could try "/usr/local/bin/guix pull"
<raghavgururajan>Noclip, On Guix system? That directory does on exist.
<Noclip>It's also just a symlink.
<Noclip>Does that one exist?: /var/guix/profiles/per-user/root/current-guix/bin/guix
<Noclip>You only need any working guix binary (doesn't has to be the newest version) for it.
<clodeindustrie>hi, I'm trying to build my first package, going a bit hands on about it
<clodeindustrie>I have to add dependencies with (propagated-inputs....) but when I add them I get a
<clodeindustrie>"unbound variables: ~S" error when "guix package --install-from-file=my.scm"
<clodeindustrie>is there anything in particular I need to add to my #use-module?
<Noclip>raghavgururajan: You should be able to find all guix binarys in your store with the following command:
<Noclip>ls -Alh /gnu/store/ | grep "^-.*-guix-command$"
<clodeindustrie>y package def https://paste.debian.net/1158718/
<roptat>raghavgururajan, "ln -sv /var/guix/profiles/per-user/${USER}/current-guix ~/.config/guix/current" should repair it
<Noclip>roptat: Oh, that's much more straight forwards xD
<pkill9>how do you get a %package-module-path with your extra channels?
<pkill9>it's from the gnu/packages.scm module, and in `guix repl` it has extra channels, but in guile it just has /run/current-system.. and "gnu/packages"
<apteryx>does someone know libtool well? I'm trying to understand why, to build X which requires to be linked with Y, it wants the shared libraries that Y depends on.
<apteryx>Y has a pkg-config .pc file which defines the libraries it links against in a Libs: entry
<apteryx>the result looks like: ld: cannot find -lsrtp2, because the X package doesn't have access to Y's shared libraries (that usually doesn't matter, as X doesn't need to directly link to Y's libraries).
*apteryx is confused
<apteryx>perhaps that'd be expected if it's trying to create a static archive? (it'd need the closure of all the libraries involved)
<raghavgururajan>Noclip,roptat: Thanks!
*brettgilio is always confused
<bobbytables54>So I'm trying out guix on virtualbox and I installed it just fine but after I reboot it takes a long time to get to gdm (assume that is the login manager) then I login and the screen goes blank. I chose the xfce desktop environment. Not sure where to start to look for errors
<apteryx> perhaps go to tty2 with Control-Alt-F2 and see if you can login a shell session
<apteryx>from there you could look at the logs such as /var/log/Xorg.0 or /var/log/messages
<bobbytables54>hmm, I think my host os is overrriding me sending that
<brettgilio>bobbytables54: Can I ask why virtualbox over qemu/kvm?
<drakonis>brettgilio: if i have to hazard a guess, its what they're used to
<brettgilio>drakonis: probably so, but nevertheless
<apteryx>bobbytables54: in GNOME Boxes there's a handy shortcut to send those combo keys directly to the guest. Perhaps Virtualbox has it also?
<apteryx>(in a menu)
<apteryx>it's been a while since I used Virtualbox
<bobbytables54>I had issues trying to get virt-manager running
<bobbytables54>so I tried virtualbox since it generally "just works"
<brettgilio>bobbytables54: you can use qemu directly from the cli
<brettgilio>with or without a head
<brettgilio>both ways are fine
<bobbytables54>I'm not familiar with the cli for qemu
<apteryx>I doubt most of us are. The QEMU manpage is like 5000 lines long or something.
<brettgilio>bobbytables54: https://fosspost.org/tutorials/use-qemu-test-operating-systems-distributions
<bobbytables54>found the virtual keyboard to get to tty tho
<brettgilio>that link should explain the basics
<apteryx>I exagerated. It's *only* 3619 lines long.
<brettgilio>apteryx: *sighs of relief*
<bobbytables54>hmm looks like a modeset issue. I'll try messing with the video options of virtualbox. thanks
<bobbytables54>hey it's up and running. Guess 3d acceleration doesn't play nicely
<apteryx>bobbytables54: congrats!
<bobbytables54>thanks. now down the rabbit hole :)
<lfam>`qemu-system-x86_64 -net user -net nic,model=virtio -enable-kvm -m 1024 /path/to/guix-image`
<lfam>For anyone who'd like to try QEMU, that's a basic invocation on recent 64-bit Intel machines. It will have 1 GB RAM
<bobbytables54>thanks. I might try that. I'd much prefer qemu over vbox
<terpri__>virt-manager works well for me under guix, but seems a bit fragile, for lack of a better term
<terpri__>for guix hackery i just use the "guix system"-generated qemu incantations
<lfam>You might need more RAM (or less!), but that's just an example
<terpri__>i'm probably going to try setting up pcie passthrough next, to run a proprietary OS under virtualization for $work, but not looking forward to it
<lfam>Time to go spelunking in the manual
<terpri__>incidentally, after weeks of discussion, $client has agreed to liberate at least the core part of their product under a free software license!
<bobbytables54>nice
<terpri__>probably to be followed by me porting it to godot instead of a proprietary game engine
<terpri__>worth dealing with C# for that ;)
***terpri__ is now known as terpri
<bobbytables54>yeah, I'm stuck with a proprietary game engine for my day job too
<lfam>Wow, that's really cool! Congrats
<bobbytables54>what the libre version of firefox? icecat? seems to be quite out of date, is there a more up to date version?
<terpri>bobbytables54, it's based on the ESR firefox version, generally. the simplest way to run regular firefox on guix is to use flatpak
<bobbytables54>ah
<terpri>i use https://firefox-flatpak.mojefedora.cz/ to install the dev edition, i believe mozilla is officially supporting firefox-on-flatpak now via flathub's beta channel: https://bugzilla.mozilla.org/show_bug.cgi?id=1441922#c59
<terpri>(obligatory warning: flathub distributes both free and proprietary software packages)
<pkill9>nice terpri
<pkill9>what kind of game is it?
<bobbytables54>yeah, I was hoping for a libre version of firefox that was up to date. I will use icecat tho, it's 10 versions behind but it will do
<apteryx>does someone know what component in the gnu-build-system is responsible for setting the runpaths on the compiled binaries?
<apteryx>I have shared objects that fail the validate-runpath phase and I'm wondering why.
<terpri>pkill9, a library for game design rather than a game itself, http://terpri.org/mgdoc/ is a pre-alpha of the manual that's basically a components list
<apteryx>bobbytables54: it uses the ESR (which is like the Firefox LTS)
<bobbytables54>yeah, I know the esr. It's just pretty old at this point
<terpri>(made a messy sed + awk + elisp script to extract the internal "HelpInfo" variables describing what the components do, very incomplete)
<terpri>there will be real games to serve as demos that i hope to make free also, but we'll see
<terpri>(i imagine "make the game source gplv3 and give ourselves a license exception to sell on terrible app stores" would be an acceptable compromise, then can become really free in practice after the supporting libraries are ported to godot)
<terpri>then it can*
<blackbeard[m]>terpri: I think they don't even need the explici exception if they hold the copyright they con probably fill a copyright complain with the store and any competition will be taken down
<apteryx>bobbytables54: You could help port the Icecat scripts to newer Firefox release and have it in a bleeding edge icecat channel if you have the motivation/time :-)
<bobbytables54>i probably will if i decide to stick with this
<terpri>most promising demo so far is an action-platformer with a protagonist who hauls around a kite on his back (and presumably has a sword or something too)
<terpri>blackbeard[m], yeah, that's true i guess. self-exception not really necessary
<terpri>i've been pushing the "selling license exceptions" model hard so it's been on my mind :)
<terpri>maybe i'll get permission to post the in-progress artwork, some of it's really cool
<terpri>also i'm quite sure i can get $client to agree to at least a free engine, proprietary assets model that the fsf has endorsed
<terpri>(iirc)
<terpri>though i prefer 100% free software & hardware & culture myself
<KE0VVT>terpri: Some art serves a functional role in the game and may need to be free.
<terpri>sure
<terpri>arguably one should provide at least placeholder assets to make the game playable, if not necessarily pretty
<brettgilio>Goodnight, all!
<apteryx>good night!
<bobbytables54>so I'm trying to compile my project and for some reason it's trying to use cc instead of gcc like the makefile specifies. Is there a place that is preferred as a pastebin here?
<lfam>bobbytables54: Yes, we prefer <https://paste.debian.net>
<bobbytables54>gotcha
<bobbytables54> https://paste.debian.net/1158728
<bobbytables54>not sure why it's trying to use cc when CXX is clearly defined as gcc
<bobbytables54>the other part of that is https://paste.debian.net/1158729
***lle-bout_ is now known as lle-bout
<bobbytables54>odd. So I added CC= gcc in my Makefile and it uses gcc. seems make just ignores whatever I tell it and uses whatever is in CC instead
<raghavgururajan>Hello Guix!
<raghavgururajan>After sending `guix build` process to background and if I end the SSH session, will the build process end as well?
<leoprikler>you want to nohup the process
<leoprikler>alternatively, try GNU screen
<raghavgururajan>leoprikler, nohup? What is the syntax for the command?
<leoprikler>hmm, nohup only works when starting
<raghavgururajan>Oh
<leoprikler>ah, okay, it's "disown"
<leoprikler>you can disown a job in your shell
<raghavgururajan>Yay! What is the syntax? Currently, the build process is in foreground.
<leoprikler>so first you do C-z, that SIGSTOPs it
<raghavgururajan>Okay
<leoprikler>then you do bg %N, where N is the job number (obtained via jobs) to run it in background
<leoprikler>then you do disown %N and after that you can exit
<raghavgururajan>`jobs` outputs only one line. [1]+ foo..
<raghavgururajan>So 1 is the N right?
<leoprikler>yep
<raghavgururajan>Thanks! it worked.
<raghavgururajan>before I disconnect, I still see the build messages appearing on my terminal.
<leoprikler>yes, because the output is not redirected by any of those commands
<raghavgururajan>Cool!
*raghavgururajan exits now
<leoprikler>perhaps there's some gdb trickery for that, but I don't know it
<raghavgururajan>Btw, does root own's the job now?
<leoprikler>nope, still yours
<leoprikler>it's just a process now tho
***daviid is now known as Guest85663
<aadcg>hi! I'm a beginner, so I was wondering if someone could guide me a bit. Sway just got upgrade to version 1.5. My goal is to do the steps to update the package definition. I'm doing this just to get some experience. I'm sure you guys can do that it easily.
<leoprikler>Package updates should be rather easy. Under normal circumstances, you only have to update the version and hash fields.
<aadcg>leoprikler: I'm struggling about figuring out the hash part. I need to use guix download?
<leoprikler>If it's a normal tarball, yes
<leoprikler>If it's a git reference, use guix hash -rx on a local checkout
<aadcg>leoprikler: oh I see. thanks!
<aadcg>is there a way to generate a patch from magit and send it with gnus?
<mroh>aadcg: try to create a patch with "Wcc" in magit and send it with the help of the emacs-gitpatch package.
<aadcg>mroh: thanks!
<aadcg>I've sent a patch to bump sway to version 1.5.
<NieDzejkob>(there's also the method of letting guix download the file once and copying the correct hash from the error about hash mismatch :P)
<alextee[m]>^ this is the correct method
<efraim>^ this is the common method, 'guix hash -rx' is the correct way
<PotentialUser-11>after installing and running guix pull guix package -u says failed to install locale. is interrupting a guix command that dangerous
<PotentialUser-11>by installing i mean installing the guix system
<guix-vits>PotentialUser-11: IDK, but: recently i've updated mine. Get some error. Repeated the command. No error. Probably it's fine (sometimes, after some big updates, reboot helps).
<PotentialUser-11>it didnt go away after reboot
<PotentialUser-11>i will try guix pull again
<nckx>Goood morning Guix!
<sneek>nckx, you have 1 message!
<sneek>nckx, raghavgururajan says: If I do `./pre-inst-env guix build foobar --root=foo/bar`, will the dependencies of foobar that are built, be conserved in gcroot as well?
<guix-vits>Hello nckx!\
*guix-vits woof woof
<nckx>raghavgururajan: Depends. Define dependencies. References, yes, build-time inputs: maybe, depending on your guix-service-type configuration.
*nckx guix-vitssnack.
<PotentialUser-11>guix pull says there is nothing to do but guix package -u exits after saying failed to install locale
<nckx>PotentialUser-11: Have you reconfigured after pulling?
<guix-vits>:)
<PotentialUser-11>ohh
<nckx>Depending on when you see it, it's possible that this message comes from the daemon.
<PotentialUser-11>what is the command for reconfigure
<nckx>sudo guix system reconfigure /etc/<your system.scm>
<nckx>PotentialUser-11: To answer you other question: interrupting guix is fine.
<PotentialUser-11>ok its updating now thanks
<nckx>You might have to ‘herd restart guix-daemon’ or even restart. If you then still get that error let us know.
<PotentialUser-11>i just installed so rebooting is fine
<nckx>Welcome, by the way.
<PotentialUser-11>thanks
<PotentialUser-11>i am sad that my nvidia graphics card is almost useless in a libre distro
<nckx>So are we. ☹
<guix-vits>+1. By itself even my mobile card is great.
<PotentialUser-11>so is intel uhd 630
<pkill9>things look a lot better after installing all the font-* packages
<PotentialUser-11>guix system: error: exception caught while executing 'start' on service 'file-system-/sys/kernel/debug':
<PotentialUser-11>Throw to key `match-error' with args `("match" "no matching pattern" ("none" "/sys/kernel/debug" "debugfs" () #f #f #f))'.
<PotentialUser-11>when upgreading shepard services
<nckx>Should be fine after a reboot. This file-system-/sys/kernel/debug is *very* new.
<PotentialUser-11>ok
<PotentialUser-11>the error is gone thanks
<nckx>👍
<guix-vits>PotentialUser-11: The error i'd get, btw.
<raghavgururajan>nckx: Okay!
<PotentialUser-11>i will get an acutal nickname
<LibreCat>this is my new nickname
<nckx>\o/
***guix-vits is now known as LibreCat
*LibreCat evil muhahaha
<LibreCat>Hi raghavgururajan.
<LibreCat>sneek later tell guix-vits: i own you.
<sneek>Will do.
***LibreCat is now known as guix-vits
<guix-vits>Hi me
<sneek>guix-vits, you have 1 message!
<sneek>guix-vits, LibreCat says: i own you.
<guix-vits>Yessir.
<emacsen>Hi all. I have a (hopefully) simple question. What is the process by which configuration in guile gets turned into config files, and should that also apply to user settings?
<leoprikler>you mean for the various service types? There's code written for that
<leoprikler>there's no universal solution, you just generate a bunch of files using the various primitives that guix offers (such as mixed-text-file)
<emacsen>leoprikler, So, here's what I'm thinking... I have a set of scripts that I run on new systems that does stuff like set gnome settings on and off. And I'm wondering if that's something that should be expressed this way
<leoprikler>if it only affects your user settings, just run those scripts
<leoprikler>if you need system-wide stuff, you should express it as service
<emacsen>I haven't tried setting, eg my printer yet.
<leoprikler>printer is cups-service-type
<emacsen>leoprikler, is there a guide beyond the reference manual. I feel like I'm not seeing the big picture
<leoprikler>There is a cookbook
<emacsen>leoprikler, well I mean even before this... the init system, the package system, what is a user profile vs system,
<leoprikler>there are some papers on that
<leoprikler>not really sure if scientific literature is your cup of tea
<pkill9>is there a way to get all the commit messages from a git repository, without checking it out?
<pkill9>and without relying on a frontend, e.g. cgit
<NieDzejkob>There's this paper by ludo himself: https://en.wikisource.org/wiki/Functional_Package_Management_with_Guix
<NieDzejkob>pkill9: I think the underlying git protocol supports this, but I'm not sure whether there's a tool/command for that
<pkill9>hmm
<pkill9>NieDzejkob: any idea off the top of your head how to get it then? with curl maybe?
<emacsen>NieDzejkob, thanks
<kmicu>Hi emacsen: it’s your choice really. By default only system is managed by Guix (System) but with additional work *everything* can be managed by it https://framagit.org/tyreunom/guix-home-manager
<kmicu>emacsen: I know at least two folks pushing idea to the extreme; Graham is one of them https://grahamc.com/blog/erase-your-darlings
<emacsen>kmicu, thanks :)
<guix-vits>NieDzejkob: thanks for the link.
<NieDzejkob>pkill9: "huh, this might be a nice side-project" ;P
<rhj>I've been having some trouble editing /etc/hosts to add a blacklist. It seems Guix rewrites it on reboot. Is there any way to make it not do that?
<rekado>I just noticed two odd things: 1) while running jackd, the Gnome Wifi settings wouldn’t come up because they were waiting for a sound to be played. 2) when clicking on the little gear next to my selected Wifi network I’m asked for a password to change settings, but when I dismiss the password prompt I get access to the settings anyway.
<bavier[m]1>rhj: use the system configuration to adjust it
<rekado>rhj: no. You would do that in your operating-system configuration and reconfigur.
<rekado>*reconfigure
<rhj>bavier[m]1, rekado: does the hosts file need to be specified as an argument, or is simply replacing it and then reconfiguring config.scm enough?
<NieDzejkob>rhj: reconfiguring will set the host file to the contents specified by the configuration
<rhj>if it's in the default location, does it need to be specified?
<rhj>or would reconfiguration make the current file the new default?
<roptat>rhj, the config specifies the content of the hosts file, and guix writes that content to /etc/hosts at boot
<roptat>not the other way around
<roptat> https://guix.gnu.org/manual/devel/en/html_node/operating_002dsystem-Reference.html#operating_002dsystem-Reference has a small section for "hosts-file"
<rhj>roptat: so would I need to maintain a separate hosts file to direct Guix to copy to /etc/hosts on reconfiguration?
<NieDzejkob>you could also put it inline into the scheme file
<NieDzejkob>in general, things in /etc are managed by Guix, not you directly
<NieDzejkob>/etc/config.scm is kinda an exception, but it doesn't have to be there. My config is in ~/.dotfiles/guix/system.scm, so that I can keep it in version control with Git
<rhj>NieDzejkob: at 9mb it would definitely have to be separate
<NieDzejkob>oh, that's the scale we're talking about. hmm.
<NieDzejkob>perhaps a blacklist like this one would work best as a package or service?
<rhj>NieDzejkob: that would be way beyond my abilities. At the moment I'm reconfiguring, and I'll see whether things work out with this setup. Thanks for the tips, all
<NieDzejkob>okay. Out of curiosity, what is this blacklist you're using?
<rhj> https://github.com/mitchellkrogza/Ultimate.Hosts.Blacklist
<rhj>might be overly ambitious of me, but we'll see
<raghavgururajan>LibreCat: Hi o/
<raghavgururajan>sneek, later tell LibreCat: Sorry, I missed your ping.
<sneek>Will do.
<rhj>Well, that didn't work.
<rhj>\builder for `/gnu/store/vrhppwm7a419gsp0ppq020zi9jw7h38z-etc.drv' failed with exit code 1
<rhj>build of /gnu/store/vrhppwm7a419gsp0ppq020zi9jw7h38z-etc.drv failed
<rhj>
<rhj>That was with (hosts-file "/home/name/.guix_configs/hosts") just under the hostname in config.scm
<alextee[m]>sorry to ask about trademarks again, but i revised the zrythm trademark policy a bit to allow modifications by free software distros approved by the gnu project:
<alextee[m]> https://paste.debian.net/1158831/
<alextee[m]>so no explicit permission is needed by guix. what do you think ^ ?
<alextee[m]>for guix*
<NieDzejkob>rhj: try (hosts-file (local-file "/home/..."))
<rhj>NieDzejkob: thanks, that seems to have got me over the hill.
***terpri_ is now known as terpri
<smileyface>Hi. I have a newly installed Guix system. What is a good way to manage installing packages defined in one or more .scm files?
<leoprikler>alextee[m]: So free Linux distros like Gentoo may patch it, but Debian and Gentoo may not? Probably want to revise that to make it a bit more inclusive ;)
<alextee[m]>leoprikler: on it
<alextee[m]>leoprikler: https://paste.debian.net/1158836/ how about this?
<alextee[m]>i don't want someone to publish it in the windows store for example :P
<alextee[m]>(or any platform with DRM, etc.)
<leoprikler>I'm not quite sure how that wording changes anything tbh.
<pkill9>smileyface: what do you mean? if you want to install packages from .scm files, run `guix package -f file.scm`
<alextee[m]>leoprikler: i think distros like ubuntu and arch would have no problem getting written permission
<alextee[m]>what matters is that free distros can patch it freely without hassles
<alextee[m]>if the distro is not free, well, sucks for them :P
<smileyface>Sorry my intent was a bit unclear. What I mean is that instead of running "guix package -i <package names>" I would like to define e.g. all my packages as a list in one file - so "emacs icecat ..." and then use Guix to install packages included in that list, so that my system goes from 1 state to the next.
<leoprikler>I just noticed a typo in my first reply "free distros like Gentoo" should be "like Guix"
<alextee[m]>my mind automatically replaced that :-)
<str1ngs>sneek: later tell guix-vits the bug with the infrastructure link on https://ask.fedoraproject.org/. Happens because requests for new windows by JavaScript was not being handled by 'decide-policy signal. It should work if you test from git. Thanks for reporting this. I'll get all of these fixes into a point release soon as well.
<sneek>Okay.
<leoprikler>I still think, you might have a bit of trouble with distro forks
<leoprikler>Like, let's assume Debian gets your written permission
<leoprikler>Does this permission then transitively apply to Ubuntu, Mint and SteamOS?
<alextee[m]>i guess the permission would be something that allows redistribution of unmodified debian, so forks would be OK, but modified forks would have to remove it or something?
<alextee[m]>how does firefox do it?
<leoprikler>it does not, that's why we have icecat
<alextee[m]>welp, tbh as long as it can get into free distros i dont care about popular distros, since i publish binaries for them anyway
<alextee[m]>it's gives my binaries more value
<alextee[m]>so i guess sticking to the FSDG wording should be fine
<leoprikler>I do think your first wording is more clear if that's your goal
<alextee[m]>leoprikler: you mean fsf-approved distros?
<alextee[m]>im not sure if that works with modified FSDG-compliant forks
<alextee[m]>the FSDG thing feels a bit more lenient but still won't allow something like publishing in the microsoft store, but im not a lawyer
<leoprikler>Perhaps, but how do you prove FSDG compliance in front of a court? Is "I'm FSF approved" a reasonable shortcut for that?
<alextee[m]>well it explicitly says "no DRM is permitted" in the FSDG
<alextee[m]>i could add a "OR FSF-approved distros" in there just to be safer
<leoprikler>I think before writing more and more, reading is a bit more important
<leoprikler>Also "no DRM permitted" is already a part of GPLv3 IIRC
<smileyface>pkill9: looking through the manual it seems the word I'm looking for is "manifest"
<alextee[m]>yeah but the gpl doesn't have something to do with the system it's running on
<alextee[m]>this is about the distribution that's publishing it, not the modified version of the software
<alextee[m]>welp, for starters i'll go with a "FSF-approved distros" wording then i guess
<alextee[m]>until someone comes up with a nicely worded more permissive wording
<alextee[m]>"nicely worded wording"
<bavier[m]1>imo, maybe not the place to discuss, but that still seems like a non-free license
<alextee[m]>bavier: it's not a license, it's a trademark policy. if you remove the logo and change the name, you can do whatever the AGPL allows you to
<alextee[m]>both the AGPL and the FSDG allow this
<smileyface>So to add on to my question from earlier: How do people manage package installation on their system? Define everything in one/more file(s), "guix package -i <package>" or some other approach?
<bavier[m]1>ah, my bad, I missed the distinction. alextee thanks for clarifying. I've seen too many gratuitous licenses with similar wording...
<NieDzejkob>I declare manifests, split across some files
<NieDzejkob>but that's mainly because I want to create containers that share a subset of the packages
<leoprikler>I personally have split manifests, that I combine into a single profile through some very arcane usage of find.
<smileyface>I would like to have split manifests too, so that I can separate packages used for work and personal usage. Other than definining the manifest files, how do you actually run the "upgrade command"?
<pkill9>smileyface: i put desktop packages in a separate profile, and then just run `guix package -i ...` for other packages
<pkill9>that's how i manage my packages
<pkill9>you can get a list of installed packages with `guix package -I` (installed to default profile)
<str1ngs>I use a monolithic manifest with lists for groups of packages. some are programmable installed depending on the system I'm using.
<smileyface>So following the example from the manual it would be enough to specify several lists similar the following? (specifications->manifest '("emacs" "guile@2.2" "guile@2.2:debug"))
<str1ngs>smileyface: that's what I do yes. here my main profile. http://paste.debian.net/1158838
<apteryx>alextee[m]: eh, thanks for your effort in trying to get your trademarks policy more friendly toward free software distros!
<alextee[m]>apteryx: 👍️ :-)
<alextee[m]>i just pushed it to git, will send a new zrythm patch in a few days when the new version is out
<smileyface>Thank you all for your advice :)
<apteryx>neat
<str1ngs>is anyone else getting guix pull: error: Git error: cannot locate local branch 'origin/keyring' with guix pull?
<str1ngs>I think I fixed it with rm -rf ~/.cache/guix/checkouts/
<str1ngs>nope still getting guix pull: error: Git error: cannot locate local branch 'origin/keyring'
<efraim>I went over the top and put all my manifests in one file https://gitlab.com/Efraim/guix-config/-/blob/master/Guix_manifest.scm
<smileyface>efraim: Thank you!
<smileyface>Can I make Guix use tmpfs for tmpfiles during compilation similar to Gentoo? ( https://wiki.gentoo.org/wiki/Portage_TMPDIR_on_tmpfs )
<apteryx>you'd just define a mount point from none to /tmp in your operating system declaration, IIUC.
<smileyface>Is it likely that some applications would use up the space given there (say 10 GB)?
<smileyface>Other than compilation of large applications
<apteryx>I don't think so, but the build artifacts kept with --keep-failed will accumulate until the next reboot
<apteryx>I sometimes clear them up with rm -rf /tmp/guix-build*
<smileyface>I see. I will try to just have a 10-12 GB tmpfs mounted on /tmp and keep that in the back of my head.
<NieDzejkob>str1ngs: git fetch origin keyring:keyring
<NieDzejkob>(that's with a local fork of the repo, right?)
<str1ngs>NieDzejkob: it's working now. so maybe it was a temporary server issue.
<LibreCat>how can i build xonotic from source
<sneek>LibreCat, you have 1 message!
<sneek>LibreCat, raghavgururajan says: Sorry, I missed your ping.
<nckx>raghavgururajan: That was guix-vits having a joke with /nick…
<nckx>LibreCat: guix environment xonotic -- true (so you don't build all of its dependencies from source), then guix build/install xonotic --no-substitutes.
<LibreCat>i meant the bleeding edge gitlab version but thanks
<LibreCat>where are the build settings stored
<LibreCat>in guix
<nckx>What are build settings?
<LibreCat>like Gentoo's make.conf
<nckx>Our source for xonotic is in gnu/packages/games.scm .
<nckx>I don't know what that is.
<LibreCat>the options for the c compiler
<nckx>You'd add them to a custom xonotic package that inherits https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/games.scm#n7269
<nckx>There is no central CFLAGS.
<LibreCat>oh ok
<LibreCat>:) thanks guys
<bavier[m]1>LibreCat: or 'guix build' has '--with-branch', '--with-commit', etc flags that you might be able to use, if the package recipe otherwise doesn't need changes
<nckx>bavier[m]1: Xonotic expects a zip.
<LibreCat>another question about building from source how can i use multicore compiling
<nckx>Wow that package can do with some refactoring.
<bavier[m]1>then '--with-source=...' ?
<nckx>git clone && zip && sure I guess.
<nckx>LibreCat: guix build/install/… --cores=n for building with -jn, and/or --max-jobs=n for building n multiple packages in parallel.
<LibreCat>how can i change -march
<nckx>Probably by adding "CFLAGS=-march=native" to the #:configure-flags. You'll have to use substitute-keyword-arguments for that if you're using inheritance, plenty of examples in Guix. Or you can just copy the Guix version wholesale & edit it.
*nckx AFK.
<LibreCat>i decided to use binary version but thank you for all of your dedication
<LibreCat>is mpeg4-aac decoder propritary software
<terpri>LibreCat, not necessarily, there's ffmpeg etc., though there may be patent issues
<terpri>guix mpv has no problem with mp4-aac
<lfam>Is mpeg4-aac a program?
<lfam>Or a standard?
<lfam>If you need to use AAC with free software, you can use the libfdk library
<lfam>terpri: Does our mpv support AAC?
<terpri>i think LibreCat is using it in a generic sense, can't easily find a problem called "mpeg4-aac"
<terpri>yes
<lfam>That's interesting, I wonder what implementation it uses. I didn't realize we had one besides libfdk
<lfam>I guess there is a decoder somewhere, and I was thinking of encoders
<terpri>e.g. from a youtube-dl'd (probably) mp4: https://paste.debian.net/1158860/
<terpri>mpv uses ffmpeg
<terpri>video.scm has the full list of configure flags (and ones intentionally left off i guess?), x264 and x265 are both enabled, along with the fully-open codecs (xiph.org) one would obviously expect
<terpri>configure flags for ffmpeg*
<LibreCat>mpv works
<pkill9>does anyone recognise this error? error: Wrong type argument in position 1 (expecting pair): ~S
<pkill9>I'm getting it often when running `guix repl <filename>`, and now I got it when running `guix build -f <filename>`
<lxsameer>hey folks, as far as i remember there was a list of open hardwares or hardwares with open source drivers and suppert by guix SD, but can't find it
<lxsameer>could you please guide me to it
<rekado>pkill9: are you using guix from a git checkout?
<LibreCat>lxsameer : https://libreplanet.org/wiki/Group:Hardware/ReverseEngineering
<rekado>pkill9: if so: run “make clean-go”
<lxsameer>cheers
<bdju> https://lib.openmpt.org/libopenmpt/license/ would this be allowed in guix or does the third bullet point break freedom 0?
<lfam>FFmpeg has it's own AAC decoder implementation: http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/aacdec.c;h=c0364033d645efe7b60f7223868ca27b90f797f7;hb=HEAD
<pkill9>rekado: no I'm not
<lfam>Makes sense
<LibreCat>its off-topic but what is the most powerful 100% libre hardware
<pkill9>probably a librebooted x200
<LibreCat>what about desktop motherboards
<pkill9>i don't know if such a thing exists
<LibreCat>there is a gigabyte motherboard
<str1ngs>some power motherboards are very open
<lxsameer>LibreCat: thanks for the list, but it's not the same as the one i was looking for
<LibreCat>what is the most powerful gpu then
<lfam>LibreCat: It's the ASUS KGPE-D16
<lfam>That's the most powerful "libre" motherboard
<LibreCat>more powerful than talos 2
<lfam>Unless you want to use an obscure architectures, in which case you can use IBM POWER
<lfam>Right
<lfam>You won't get anything faster than that AFAIK
<LibreCat>how can i fit that in a chasis
<LibreCat>and what is the TDP
<LibreCat>ok but what is the most powerful GPU that is 100% free
<pkill9>rekado: any idea what could be causing the issue?
<stikonas>LibreCat: I guess intel integrated GPUs
<stikonas>definitely more powerful than GPUs targeting ARM
<pkill9>rekado: I narrowed it down to this line: ,@(alist-delete (package-inputs wlroots) "wayland")
<pkill9>in an inputs specification
<pkill9>i think I had the arguments the wrong way round
<pkill9>yes that was the reason
<bdju>LibreCat: are you talking runs libre software / firmware or the hardware itself is libre? (with designs and such you can modify) The MNT Reform might fit the bill. The schematics are all online for the keyboard, trackball, case, etc.
<LibreCat>i mean no firmware driver
<lfam>The MNT Reform is far from powerful
<lfam>It's definitely a cool project but it's based on a cell-phone SOC
<rekado>in my old workstation (assembled when the AMD Phenom II was new) I have an Nvidia Quadro 600 GPU card.
<joshaspinall[m]>some of the older Nvidia cards work very well with the nouveau driver, and they seem to be catching up to the newer cards quickly looking at theirfeature Matrix. Sadly most AMD cards require proprietary Firmware to be loaded, despite using FOSS Drivers
<rekado>this card works (i.e. I can use Blender’s sculpt mode without annoying lag) with Linux libre.
<rekado>almost 10 years old now :-/
<rekado>but I guess it really depends on what you need it for
<rekado>unfortunately, information about what GPUs will work with Linux libre is hard to come by.
<rekado>h-node has only very few records for GPUs.
<christianbundy>Hey all, anyone know if newer versions of node are going to get packaged?
<Formbi>there are some issues with bootstrapping
<Formbi>I made a node package that pretty much works, but due to said issues it's not really ok in Guix
<hendursaga>I take it something MIT-licensed won't/can't be part of the main Guix package repo?
<lfam>hendursaga: The MIT license is a free software license. Guix has probably thousands of MIT licensed packages
<hendursaga>lfam: Then how come 'guix licenses' doesn't show it?
<lfam>In Guix, the MIT license is called "expat", because there are actually a few different licenses that have called themselves MIT.
<leoprikler>because there's different mit-style licenses, expat and x11 come to mind
<lfam>Right
<hendursaga>Oh! That's why I keep seeing it, haha
<lfam>It's a shame the situation is so confusing
<Formbi>hendursaga: nope
<Formbi>it's because of some machine-generated code
<kmicu>We could safely recognize that today mit = expat and move on like rest of the world.
<Formbi>christianbundy: for now you can get the not-so-nice Node 14.6 here: https://gitlab.com/Formbi/my-guix
<lle-bout>kmicu: legally it's important to be unambiguous
<kmicu>Text is unambiguous. That’s what is legally important.
<bavier[m]1>so, we give every package "fsf-free" license and point to the source?
<kmicu>That’s slippery slope argument. We could just alias expact as mit and recognize that’s what folks today want when searching for mit.
<bavier[m]1>that seems to be what spdx.org does, while recognizing some "common" variants. they give "x11" it's own identifier.
<bavier[m]1>but this would be better discussed in #fsf, I think
<kmicu>That’s what SPDX, OSI, Github, Gitlab, choosealicense do today (I have string reasons to not like those organizations) but nowdays FSF’s expat/x11 distinction instead preventing confusion just creats one. Oh sweet irony.
<lfam>I think we can leave it alone
<kmicu>That’s what Emacs did with its frames and windows distinction.
<Formbi>or maybe it could be called mit-expat and mit-x11?
<lfam>Just because we can do something, it doesn't mean that we should
<Formbi>Emacs didn't /do/ it
<Formbi>it had it since a long time ago
<lfam>In practice, it's not a serious barrier for people making Guix packages
<kmicu>From my perspective it’s unnecessary confusion but I get your point on mantaining status quo and tradition. (That doesn’t change the fact that world move on and mit = expat.)
<kmicu>Formbi: Emacs predates X11 I understand why it uses that terminology, it’s sweet, but that terminology is nevertheless outdated.
<terpri>i'd have a hard time arguing it's a "serious" barrier, but it a barrier
<leoprikler>maybe declare license:mit as (error "Did you mean license:expat or license:x11?")
<Formbi>I can kinda agree about changing the frame/window thing
<Formbi>but some people also would like Meta to be called Alt and shit like that
<terpri>or maybe declare license:mit as an alias for license:expat which is the terminology almost everyone else seems to use
<leoprikler>key here is "almost"
***Guest85663 is now known as daviid
<terpri>"almost" might be "everyone except fsf/gnu", although debian says mit/expat in places