<pkill9>although technically i guess you could yea
<pkill9>but i was just thinking it would be reliable, you wouldn't just have one file for the latest guix, you couldn't have one for each revision, but it's just a boring detail i was thinking of
<pkill9>also i dunno if all applications with desktop files generate an appstream file
<nckx>Ah, I didn't mean to be pedantic, those were genuine but confused questions since I'm extremely out of my depth with this GUI stuff. I've never used the kind of thing at which this file is obviously aimed.
<pkill9>this is why i just want to be able to get all the desktop files, although i suppose you could fill in missing appstream files with metadata from the .desktop files
<nckx>It doesn't seem that common; I have 25 unique ones on my system. But I don't use GNOME of course.
<Sleep_Walker>is there a way, how to specify user's initial set of packages via system config? so when I init new system via `guix system init`, the packages are in /gnu/store ready, user can have them installed without any network requirements and it's not linked to system profile?
<leoprikler>you can specify global packages, but for everything else I'd run guix pull in the booted system
<leoprikler>Guix will automatically reuse packages, that have already been built
<vagrantc>you could define a package that installs nothing itself, but has inputs for other packages and include that in the system profile ...
<vagrantc>then all the packages you want would be readily available in the store without polluting the global install...
<vagrantc>something like that, if not that exactly
<nckx>Sleep_Walker: Check out gc-root-service-type. It's used by (gnu system install) to reduce downloads during installation.
<tricon>Just wish to continue to extol my adoration for Guix. Ya'll are doing a great job.
<str1ngs>I know right, for some reason my Linux never needs restarting after an update. something must be wrong!
<brendyyn>str1ngs: i heard it has something to do with how the filesystem works differently between unix and windows. on unix when you update a file, programs reading it at the time still refer to the old version. where as windows locks files so you gotta stop and restart with the new file
<brendyyn>str1ngs: many things do also break after updating on GNU/Linux anyway so i think its not quite right to tell users they can just update and keep going. i actually think it may be desirable to add a may-require-restart property to packages known to break on update to notify the user when they are updated
<str1ngs>brendyyn: I don't find things break on upgrades at all.
<raghavgururajan>nckx: I have a doubt. If a package declares a env-var, will that env-var be available right after install phase inside build-env? For example, let assume package foo supposedly establish/set an env-var called FOO_PATH. (like how elogind sets XDG_RUNTIME_DIR). Now, if I wrapping execulatbles of package foo, via (getenv FOO_PATH), right after install phase, will getenv receives any value?
<fnstudio>wigust: that's very helpful, thanks, and i think it's exactly the problem with my setup
<fnstudio>wigust: any idea on why my `xdg-mime default ...` edits `~/.config/mimeapps.list` instead?
<wigust>fnstudio: probably because this file should be for MIME configuration by conventions ;-) I didn't find a documentation when I wrote my config and didn't strace update-desktop-database :-)
<fnstudio>wigust: i see, makes sense, thanks a lot for your help
<brendyyn>str1ngs: I don't know about guix but I've found KDE does get a bit funky.
<aadcg>I'm trying to install a package with rpm on guix with sudo rpm -i file.rpm and I get -> error: cannot open Packages database in /gnu/store/5ci625m65m8jpv8xiq2wna3s1rmj94x2-rpm-188.8.131.52/var/lib/rpm. Any ideas?
<nckx>raghavgururajan: So I don't understand your question about variables. I thought you meant [native-]search-paths, but elogind doesn't declare one. What does ‘establish/set an env-var’ mean, *exactly*?
<nckx>raghavgururajan, str1ngs: The only bean of which I approve is the fruit of the C. arabica.
<nckx>aadcg: So... that's not going to work. rpm is packaged in Guix to allow inspecting/unpacking/&c. .rpm files, but it is not remotely set up to install packages, and Guix System is not remotely set up to accept them (they would assume tonnes of things about the system, like FHS, that just aren't there).
<nckx>alextee[m]: Does the meson-build-system not accept a #:meson or similar argument where you could specify a newer meson? Then you could create a meson-0.55 package on master, use it only for zrythm without causing rebuilds.
<nckx>Of course a global meson update on c-u would still be very welcome.
<str1ngs>pkill9 view is just a Nomad term for presenting a webpage transparently. it's essentially just sxml to html strings. here are other moving parts that you don't see in (nomad views) mainly some webkit magic. though it's generic enough I hope
<str1ngs>joshuaBP` also nomad is in guix :). though its still pretty early days.
<joshuaBP`>str1ngs: it's still pretty freaking awesome!
<butterypancake>wigust: Patch #42674 had its prerequisite patch applied if you'd like to review it again. No pressure, just thought I'd let you know
<butterypancake>also I have a searx package patch I'd like to send in but I want to wait for patch #42808 first, so if anyone has some spare time, reviewing that one would make me really happy :D
<apteryx>how are debugging symbols found when using 'guix environment' with debug outputs?
<apteryx>ah, I'm using gdb from emacs which probably doesn't know about my environment profile
<butterypancake>well the debbuging symbols are pulled from the binary right? So you do `gdb -i=mi binary' and you get the symbols right?
<apteryx>but I also want the debug symbols of some library, namely qtserialbus (I've added a debug output for it)
<butterypancake>oh, a dynamic library? That's out of my comfort zone as an embedded developer :P if you want to make it easy you can compile against the library statically so the symbols are in your binary
<butterypancake>oh ok. But in this case there is just some user global gdbinit when really that gdbinit really should be per profile? is the .guix-profile/lib/debug populated by guix? So you want `guix environment blah' to add additional debug paths?
<apteryx>butterypancake: yeah, it'd be nice if there was a mechanism to automatically take care of that, but I don't know one (yet).
<butterypancake>I remeber having a .gdbinit but I removed it when I was cleaning my home folder a while ago and apparently those don't regenerate :P
<apteryx>nope, they're put there at the original installation time, and then left untouched.
<apteryx>it'd be risky to mess with those at any later time.
*apteryx investigates why guix-set-emacs-environment stopped working
<butterypancake>so looking into gdb itself, the only way to modify the debug folder locations through environment variables (which would obviously be ideal for guix profiles) is to add those folder to your PATH variable. This means we probably should handle debug symbol files in the exact same way we handle the profiles bin folder and then just add that debug folder to the path
<apteryx>or, to reproduce which your ~/.guix-profile: (guix-geiser-eval (format "(search-paths-specifications \"%s\")" (expand-file-name "~/.guix-profile/etc/profile")) (guix-get-repl-buffer 'internal))
<butterypancake>idk what your doing, but running that code blindly I got: ("()")
<butterypancake>apteryx: I'm currently on a foreign distro so probably don't mind me. I still have problems on GuixSD but they usually don't look like that
<butterypancake>usually they're something like: we don't like that guile key you sent me, but we don't like it like 3 layers down so you can't just remove the key from the repl command and have it work. I should probably debug it but I never bothered to learn guile so it'd be hard. I'll learn guile eventually
<PotentialUser-79>My ext4 filesystem always gets corrupted (2-3 times/month), requiring me to use the installer image to run fsck to repair it. It never happened before on other distributions.
<butterypancake>You run the latest Guix? You did a `guix pull' and `sudo guix system reconfigure' after installing?
<nckx>butterypancake: It's Libreboot. It's most often used with a GRUB payload which reads your operating system's grub.cfg directly, but it also has a ‘BIOS’ payload. Coreboot (on which it's based) also had a UEFI implementation but not sure if Libreboot does.
<vagrantc>sneek: later tell lfam yes, the pinebook pro patch just pulled in the device-tree, if i remember correctly, which is now in upstream as of 5.7.x
<butterypancake>I was just about to ask how you turned off your computer. Because now it sounds like the problem occurs based on the way you use it. Some distros handle hard shutdowns better than others.
<butterypancake>Shuttind down while the OS is using the hard drive is usually not a good thing to do
<PotentialUser-79>It happens when I run a guix command like guix pull, I get a "Bad message" error and when I restart my computer it fails to boot and tells me to run fsck manually.
<butterypancake>Ya, I had a computer die like two days ago from restarting after a bad message. It did it consistantly too even after a reinstall. However my 3 other machines are just fine so I though it was a hardware issue. Maybe it was a config issue. Can you post a link to your config.scm?
<nckx>(libfive/test/heightmap.cpp:261:5: error: ‘BENCHMARK’ was not declared in this scope) ☝
***bkv is now known as bqv
<butterypancake>oh ok, I guess I never understood the difference. So what I was calling booting with BIOS is actually where the machine just starts executing code right away (in this case GRUB code) where as efi does some fancy stuff first so it needs it's own fancy partition
<nckx>97% generic Guix instructions that have nothing to do with Libreboot and probably shouldn't be there, 3% not clear enough 😛
<butterypancake>someone should tell them just to take it down. It's not adding any value. I guess we could add "works with libreboot" at the top of our manual because that's all that page really seems to mean
<nckx>‘Any users, for their generalised use cases, need not stumble away from this guide to accomplish the setup. Advanced users, for deviant use cases, will have to explore outside this guide for customisation’
<nckx>Surely you're not a stumbling deviant who prefers the official docs.
<vagrantc>the author appears to be an active outreachy contributor to guix
<butterypancake>I posted respectfully in the libreboot IRC so they can deal with it as they see fit
<butterypancake>nckx: I apologize for messing up the patch thread for the openrc stuff. I've been experimenting with git send-email's --in-reply-to and chainReplyTo. I think I understand it now, but the threads for that report are not right :P emails are hard
<butterypancake>I assumed that when you send a patch series --in-reply-to something, that the entire series would be in-reply-to that. Silly me. How could I not see that the obvious outcome of that would be send the first email in-reply-to that and the second email to have no in-reply-to field. How could I be so naive
<nckx>butterypancake: Yes they are. Only people who can't be bothered to *try* to get things right need apologise; not you.
<nckx>I thought chainReplyTo would result in: <patch 1, In-Reply-To (say) the bug number> <patch 2, In-Reply-To patch 1> <p 3, I-R-T p 2> <...> etc.
<nckx>Giving something of a triangle vibe in MUAs. Instead of all patches being In-Reply-To the bug number, which is what you want.
<butterypancake>wait, the documentation agrees with you. So I do want chain-reply-to set to no, not yes. But then why did my 2/2 not get sent to the right thread?
<butterypancake>after reading the doumentation, I've come to the conclusion that there is no possible way I did what I did because subsequent emails always have an in-reply-to thing. There is no way to turn that off
<butterypancake>I'm going to assume I accidentally deleted something during annotation and that next time it'll hopefully work
<butterypancake>the weird date thing is apparently the X-Microsoft-Original-Message-ID
<nckx>I sent myself 5 patches (without chainReplyTo set) and 2-5 are all in-reply-to 1. 1 is in-reply-to a bogus message ID I gave on the command line. That's what I consider ‘normal’ (2-5 are indented under 1, sure, it's not aEsthEtic, but it doesn't break any threads).
<butterypancake>the in-reply-to of the second email matches the X-Microsoft-Original-Message-ID of the first. I think that means I did everything right and then microsoft decided to mess with my message-id just to mess with me
<butterypancake>well, I'm going to try out Guix ontop of sxmo on pinephone. If that works, I might try to get actual GuixSD working on it. Are most of the packages working on aarch64? That's really going to be my deciding factor
<vagrantc>i've got it working enough that i've run a pinebook and pinebook-pro with guix system
<vagrantc>web browsers tend to be a bit hard to come by ... netsurf is the only one i seem to be able to reliably build
<nckx>I think so. All of the ‘base OS’ at least (our aarch64 build nodes run Guix System) and I often compile random GUI things for aarch64 just as a smoke test. Most.
<vagrantc>and netsurf is really fast because it's missing support for a lot of annoying things
<butterypancake>sounds alright then. maybe learning uboot is the next thing on my agenda then. I'm not exactly sure what I'll need to do to port it, but I'll start with getting Guix on Alpine Linux to work and go from there
<apteryx>ok, so totally disabling IPv6 didn't help.
<apteryx>stopping nscd also didn't help. These addresses in the log got me thinking: 192.168.55.1 and 192.168.50.37; they don't exist in my network, these are configured through a VPN connection that I keep. It seems NetworkManager didn't bother cleaning up /etc/resolv.conf when the connection was closed/died.
<apteryx>OK, nmcli con up vpn, then nmcli con down vpn does take care of cleaning up the /etc/resolv.conf file. Not sure what happened, but my DNS latency problem is happily resolved.
***OsamaChibani[m] is now known as Chibani[m]
<apteryx>nckx: it also fixed my dockerd connection problem (that I mistakenly thougth was related to TLS certs)