IRC channel logs

2021-12-07.log

back to list of logs

<mbakke>this is new: ldconfig: Can't link /gnu/store/5sbldxhgxwn2cjlkgak8sz1xms5paw2b-mesa-21.2.5/lib/libXvMCgallium.so.1 to libXvMCr600.so.1.0.0 (from ufo2map build log)
<mbakke>not fatal, though
<civodul>mbakke: that happens during the phase that builds the ld.so cache, right?
<mbakke>civodul: indeed, I noticed it in several builds now
<civodul>mbakke: could you collect them in a bug report?
<civodul>i guess it's not fatal, but it's weird
<mbakke>civodul: sure, it seems to be just that one Mesa library though, will investigate more tomorrow
*civodul -> zZz
<civodul>alright
<civodul>g'night!
<mbakke>gn :)
<pyrotek45[m]>setting up emacs is kinda annoying
<pyrotek45[m]>when stuff doesnt work
<pyrotek45[m]>for some reason the evil-leader package just doesnt work for me
<pyrotek45[m]>at all
<the_tubular>pyrotek45[m] are you using guix home to setup emacs ?
<pyrotek45[m]>no
<pyrotek45[m]>just emacs
<apteryx>pyrotek45[m]: you have emacs and emacs-evil-leader in the same profile?
<natewrench123456>Does Guix use the Hurd Kernel?
<lfam>natewrench123456: By default Guix uses a variant of Linux called linux-libre
<lfam>It is possible to use the Hurd but it's basically experimental compared to Linux
***the_tubular50 is now known as the_tubular
<jgart>guix shell streamlink -- streamlink https://www.twitch.tv/videos/1225580773 360p --player mpv
<zoglesby>pyrotek45[m]: back to the DM topic from yesterday there is a project that uses Sway as a DM, but it is not in Guix. I have used it on other distros in the past. Maybe it is something I will look into packaging to help me learn more about Guix internals. https://sr.ht/~kennylevinsen/greetd/
<zoglesby>The guix service part being the issue, not really the package building itself.
<guixy>hi guix
<guixy>It looks like manuskript is incompatible with the current version of python-markdown.
<guixy>Manuskript fails to start when built from guix commit 05deb26be92c314b6a60ed2ca93596b895515ca6
<guixy>But it does start if I specify --with-input=python-markdown=hello
<lfam>zoglesby: There are actually pending patches for greetd
<nckx>I like how ‘hello’ is basically Guix for <none>. At least that's how I use it most. guixy: If nobody responds, could you send that info to bug-guix at gnu dot org? It's likely to get lost otherwise.
<nckx>(And thanks!)
<zoglesby>lfam: nice!
<lfam>Heh, that hello thing is very idiomatic
<zacchae[m]>Anyone have a minimal example of an email server setup? Seems like it should be as easy as pointing to SSL keys and setting domain/port, but not sure
<zacchae[m]>though actually, more generally, where do you even start to look for guix config examples outside the cookbook? I feel like I can never get search hits outside the manual (maybe this is a sign), but I see people share links all the time.
<guixy>Is hello the reason there is no officially "empty" package?
<raghavgururajan>zacchae[m]: http://search.disroot.org/search?q=guix%20config%20dotfiles&categories=general&language=en-CA
<podiki[m]>there's some on sr.ht, gitlab, and github (searching just for "guix" on there is usually enough)
<zacchae[m]>thanks raghavgururajan and podiki!
<podiki[m]>A cataloging of useful configs and general "people's guix stuff" would be nice, though for official usage would have to include free software only
<podiki[m]>while normally these things can be out of date, including a date and/or a guix describe would always allow you to look back
<zacchae[m]>yeah, it would be pretty helpful when trying to get something running to have a GARUNTEED-working starting point. Which is totally possible given the functional nature of guix
<apteryx>sneek: later tell civodul hello! re static linking to cut the dependency graph (rustc/cargo) -- at what point is it good enough? IIUC, gcc/glibc are not easily embedded; I've built llvm with 'package-with-relocatable-glibc' for example, but store references to gcc-lib and glibc remained
<sneek>Will do.
<raghavgururajan>I am hearing good things about iwd (https://iwd.wiki.kernel.org/). Should make it default instead of wpa_supplicant?
***lukedashjr is now known as luke-jr
<raghavgururajan>*we make
<the_tubular>First time I hear about this, looks interesting though
<jpoiret>hello guix
<jpoiret>does anyone know what the remaining blockers on c-u-f are?
<jpoiret>for me, there's only xdg-desktop-portal that's not working well enough, but that's also a master issue I think
<rekado_>I'm still fixing 69 bioinfo packages
<jpoiret>ouch
<zimoun>hi!
<mothacehe>hey zimoun
<zimoun>mothacehe: ah many green today for c-u-f \o/ (at least for what I track :-))
<mothacehe>hehe yeah, i rebuilt things massively
<mothacehe>we now have 90% coverage on x86_64
<ptdevnull>Is anyone else using zsh as their user shell on guixSD? I used (use-modules (gnu packages shells)) and added (shell (file-append zsh "/bin/zsh")) to my user in /etc/config.scm, and it works, but after rebooting Xfce and Enlightenment can't find any of my applications.
<ptdevnull>I'm not sure what exactly happened. Also, gdm doesn't automatically show my username anymore, but I can still type it in.
<jpoiret>do you have a .zprofile?
<jpoiret>often, GDM doesn't show your username if you don't have a password
<ptdevnull>I do, it was autoinstalled with zprezto. Is there some change I should make?
<ptdevnull>I definitely have a password. I use it to log in.
<jpoiret>yes, you need to source /etc/profile
<jpoiret>the way DMs work on guix by default is that they launch the sessions through a login shell, so that everything (env vars notably) is set up properly. Thus, you need your profile files to set up the Guix env vars, and the system's /etc/profile serves that purpose
<ptdevnull>I see, why does it go in .zprofile and not .zshrc?
<ptdevnull>This might be an elementary question, sorry.
<jpoiret>zshrc is sourced whenever you launch any shell, whereas zprofile is only for a login shell
<ptdevnull>Thanks. Hopefully this works. Time to reboot.
<zimoun>mothacehe: ah you did it. Cool! I was waiting and I was checking to check the missings. And good weather this morning. ;-)
<jpoiret>been thinking of adding a way for shepherd services to fetch values from their dependencies, mostly because for user sessions many services set up environment variables and such
<jpoiret>eg dbus-daemon creating a DBUS_SESSION_BUS, compositors creating DISPLAY/WAYLAND_DISPLAY, etc...
<jpl01>I'm just curious is there some way to create a graphical live iso with guix system disk image or something else? I know that a graphical live environment is not really needed just wanted to know if something like that would be theoretically possible.
<jpoiret>jpl01: completely, and if you don't need the installer program it's even very simple
<jpl01>alright, thanks for the info jpoiret.
<jpoiret>I suggest using `guix system image -t iso9660 yoursystem.scm` on a system definition you have lying around to try it
<jpl01>will do.
<civodul>o/
<sneek>civodul, you have 1 message!
<sneek>civodul, apteryx says: hello! re static linking to cut the dependency graph (rustc/cargo) -- at what point is it good enough? IIUC, gcc/glibc are not easily embedded; I've built llvm with 'package-with-relocatable-glibc' for example, but store references to gcc-lib and glibc remained
<civodul>sneek: later tell apteryx make-bootstrap.scm has all the tricks; one of them is to call remove-store-references to make sure that binaries actually work when run in the vacuum, such as an empty container
<sneek>Okay.
<Mahesh>Hi I'm new to guix, I'm wondering how to apply the configuration without downloading any packages, whenever I run guix system reconfigure /etc/config.scm it downloads packages every time
<rekado_>it only downloads things it needs, e.g. when you've changed the version of Guix since the last time you ran reconfigure or when you've changed the config to require more software.
<vivien>Also sometimes it needs to download again things it has just garbage collected
<pinoaffe>hi guix, I want to start a guix shell with a package whose definition was loaded from a .scm file (aka the `guix shell` analogue of `guix package --install-from-file=package.scm`), is this possible?
<civodul>pinoaffe: hi! yes, "guix shell -f package.scm" should do exactly that
<pinoaffe>civodul: oh based on the `guix shell -h` output, I thought that'd only add the package's inputs to the environments, and not the package itself
<pinoaffe>thanks!
<pinoaffe>thanks!
<civodul>pinoaffe: you'd have to do "guix shell -D -f file.scm" to achieve that :-)
<civodul>it's more consistent than "guix environment" in that regard!
<attila_lendvai>can i mutate records defined/created with define-record-type* ? more specifically, a service-configuration that i have defined for my service?
<attila_lendvai>hrm, looks like i can at least copy it, and override some fields
<civodul>attila_lendvai: yes, you'd use "inherit" to copy these records
<civodul>mutation is alien in Guix ;-)
*attila_lendvai is staring at the macroexpansion, because something doesn't seem to work
<attila_lendvai>ehh, match-record tests the object type once for each field
<pinoaffe>civodul: in that case, I think it'd be a good idea to change the --help documentation of `guix shell` and/or `guix environment`, it's somewhat confusing that `guix shell -f package.scm` and `guix environment -l package.scm` do different things even though `-f` and `-l` have the same description, word for word
<civodul>hmm yes, that's confusing
<civodul>the one in shell.scm is wrong, i'll change it
<pinoaffe>thanks :)
<civodul>the first time i read it i wasn't sure whether it was wrong, but it is
<attila_lendvai>well, fixing that in match-record was quick
<attila_lendvai>when i start qemu with `guix system vm ...`, then there's a 13 second delay in the guest vm before "Freeing unused kernel image (initmem) memory". any ideas how to avoid it? can't find much online...
<attila_lendvai>is there any way for multiline string constants (without escaping hell, similar to shell's <<EOF)? i found this, but srfi-109 is not implemented in guile: https://www.reddit.com/r/scheme/comments/atftqq/multiline_sting_literals_in_guile/
<jlicht>hello guix!
<raghavgururajan>Any thoughts on this patch-set? https://issues.guix.gnu.org/51885
<raghavgururajan>> ‎civodul‎: mutation is alien in Guix ;-)
<raghavgururajan>Made my day, civodul. :D
<rekado_>oh: sparse/tests/test_compressed.py::test_slicing[compressed_axes0-0] Fatal Python error: Segmentation fault
*rekado_ updates
<tex_milan>@raghavgururajan link to blacksmith doesn't work
<vdv>@raghavgururajan maybe sth with ice-9 pretty-print?
<vdv>(use-modules (ice-9 pretty-print)) and (lambda_ (pretty-print '(......))
<vdv>ah sry wrong one sry raghav :D
<vdv>i meant @atilla_lendvai
<vdv>@atilla_lendvai https://www.gnuvola.org/software/guile/doc/Pretty-Printing.html
<florhizome[m]>I reported that sddm has been failing to work for me recently (just doesn't start). Here a logfile: http://paste.debian.net/1222421 ...what i see is that it invokes startx despite me configuring it to run a wayland session \?/
<permcu>Hi, I am trying to do same package hacking. It is a "evil" binary package and I am using a computed-file as source. The computed-file tries to do a url download - imagine (invoke wget url) - and fails because dns looup (eg. resolve.conf) is not configured/missing. I looked into the build-in origin-methods (cvs-download.scm et al), but since two hours can not figure out how they do the dns lookup. They just seem so work. Maybe one of you
<permcu>has some pointers how the build-in download methods do dns lookup or how I can do it inside the computed-file source.
<efraim>florhizome[m]: have you configured sddm to use wayland? https://git.sr.ht/~efraim/guix-config/tree/master/item/3900XT.scm#L139
<florhizome[m]>Yup :/
<florhizome[m]>I also just checked the source and it’s not really clear to me where that „startx“ comes from.
<jpl01>Is there any precaution I could take in order to make sure that nothing malicious from a third party channels gets introduced after using guix pull?
<roptat>permcu, I think computed-file don't do network at all, so it can't resolve DNS
<roptat>permcu, you probably need to use an origin with url-fetch as the method (from guix/download.scm)
<roptat>jpl01, I don't know of any. make sure it's signed (that there is an introduction), to make sure only authorized people can make changes to that channel. But other than examining all commits yourself, I don't know what you could do. Using a channel means you trust the people who make it
<jpl01>roptat: Is there a way to examine new commits from guix itself or do i need to check every package definition on the site where the channel is hosted? Even if the channel is signed there is still a potential that someone with bad intents takes over the computer of one of the trusted committers and uses this stolen identity to push malicious changes, I just wish that there would be some way to see the changes that you're pulling from
<jpl01>guix.
<roptat>mh, there's no built-in way, but the files are there somewhere
<roptat>~/.cache/guix/checkouts contains the git repos that are used to pull
<efraim>you can check the copy of the repo guix uses
<efraim>^^ like ropat linked
<civodul>hey roptat, thanks for the guile-netlink release!
<roptat>civodul, yw :)
<permcu>Thanks roptat. I think I will have to use a custom origin-method. wget was just an example. I am actually using arduino-cli to setup an arduino environment with guix. I don't want to compile it from source, because it's huge with all the custom compilers and so forth. But I also don't want to use a docker-image or similar. Inside the custom origin-method method download works. With computed-file it does not. I can still not figure out
<permcu>why, because both use gexp->derivation under the hood. But for now I accept that only origin-method can download;)
<efraim>civodul: we might be able to drop the generate-checksums phase to near 0 seconds, turns out we don't actually have to hash the individual files
<efraim> https://wiki.debian.org/Teams/RustPackaging/Policy , paragraph containing 'cargo-checksums'
<efraim>I remember reading something about that
<civodul>efraim: nice!
<civodul>though my current concern is more about turning Rust packages into "normal" packages :-)
<efraim>I think I'm going to go through and see about locking some librsvg crates in place, it'll likely end up creating a partial rebuild on x86_64 at the end
<civodul>yeah, either that or the full-bundling approach you initially proposed
<civodul>either way, the sooner the better
<civodul>i just want to be done with this branch
<efraim>the bundling will make it be just done with it, but it's not the right approach
<efraim>it is so tempting though
<civodul>yeah
<civodul>but pinning 200+ packages?
<efraim>I bet the cycle I found earlier is something like gtk -> rust-gtk-something -> librsvg
<efraim>in some sense it's not much different than pinning pkg-config
<efraim>gtk -> rust-gtk-something -> librsvg -> gtk
<efraim>I'm not sure making them all regular packages will work once we take into account gtk stuff
<efraim>I'll see if I can get my thoughts somewhat coherent in a mailing list email
<tex_milan>If I modify scm inside ~/.cache/guix/checkouts of some channel, is it picked up by guix system reconfigure?
<civodul>tex_milan: ~/.cache/guix/checkouts is not meant to be modified by users
<civodul>efraim: whether it's #:cargo-inputs or 'inputs' shouldn't make any difference; the syntax changes, but the graph remains the same
<tex_milan>I know :) but will it be picked up? I am not doing pull for that case.
<civodul>it won't be picked up unless you run "guix pull", and if you do, the changes will be lost, i think
<efraim>civodul: the graph doesn't "see" the cycles between packages until they're moved to inputs
<efraim>until then we have our magic sauce with package-sources
<civodul>efraim: you mean there are cycles right now but we don't see them because we only look at the source?
<civodul>then you know, there's the <package> graph, and there's the bag/derivation graph
<civodul>IIUC, the <package> might be cyclic, but the bag/derivation graph isn't, right?
<luis-felipe>sneek: Later tell ngz thanks for reviewing and integrating my package :)
<sneek>Got it.
<luis-felipe>Howdy o/
<efraim>civodul: I can find one using gtk-pixbuf and librsvg on c-u-f, give me a second
<luis-felipe>Is it just me or do you see "--list-profiles" listed twice when running "guix package --help"?
<luis-felipe>(guix 51bf893)
<efraim>glib is easier to find the cycle
<civodul>luis-felipe: hola! i see it only once; could it be a bug in the translation?
<efraim>librsvg -> gtk+@3 -> inkscape -> dblatex -> gtk-doc -> glib -> rust-glib-0.9 -> librsvg
<efraim>and there's multiple paths from glib to librsvg
<roptat>luis-felipe, I see it only once
<luis-felipe>civodul: Oh, it seems like so.
<luis-felipe>roptat: Yeah, it is related to the Spanish translation then. Thanks for checking.
<roptat>luis-felipe, ah I'm on core-updates-frozen, maybe there's a difference
<roptat>luis-felipe, if you find the culprit, can you fix it on https://translate.fedoraproject.org/projects/guix/guix/es/ ?
<luis-felipe>roptat: Sure. I'll have a look.
<roptat>luis-felipe, found it: https://translate.fedoraproject.org/translate/guix/guix/es/?q=list-profiles&sort_by=-priority%2Cposition&offset=1
<roptat>I wonder why it was marked as OK
<roptat>if you feel like translating more, don't hesitate ;)
<roptat>and tell me when it's fixed, so I can push the update this evening
<luis-felipe>OK :)
<luis-felipe>Ha, no wonder why I couldn't find the option to export manifests :D
<roptat>we might have to find a way to flag identical translations, I've seen similar issues in the Russian manual
<luis-felipe>roptat: If it's just a matter of translating and saving that string, I think it's done.
<roptat>luis-felipe, thanks! it should be all I need
<roptat>I'll push it tonight
<civodul>efraim: "wtf" is what comes to mind :-)
<luis-felipe>ok, thanks.
<civodul>efraim: so right now "guix graph --path glib librsvg" succeeds, but not the other way around; that's because of the "invisible" #:cargo edges, right?
<roptat>as I said, if you want to do more work on the translation, feel free :)
<roptat>at least those would be great, if you have time: https://translate.fedoraproject.org/translate/guix/guix/es/?q=state:%3Ctranslated
<efraim>looks like that isn't the right one, the second glib is glib-with-documentation I think
<efraim>but yeah, the invisible cargo edges are there somewhere
<luis-felipe>roptat: I'll see what I can do :)
<acrow>tex_milan: I'm curious if you made headway in getting rid of that annoying BT error on desktop startup? It looks to me like that bluetooth configuration file can be easily hacked in guix/gnu/services/desktop.scm.
<acrow>The function is 'bluetooth-configuration-file'. Duh.
<tex_milan>@acrow not yet, I am working all day. Will look at it this evening again.
<tex_milan>@acrow as I installed it recently I have not yet set it up for hack ability with system reconfigure...
<rekado>I'm stumped by the build failure of bwa-pssm on core-updates-frozen
<rekado>it fails with "pssm.c:(.text+0xdeb): undefined reference to `map'"
<rekado>pssm.c has a function prototype: inline int map(const unsigned char *letters, int length, int alphabet_size);
<rekado>I don't know what would provide the implementation of this procedure.
<rekado>it builds fine on the "master" branch.
<civodul>you prolly need s/inline/extern/
<civodul>otherwise it's not quite a declaration
<civodul>./pssm.c
<rekado>thanks, I'll try that
<civodul>yeah i'd just remove "inline" both for the declaration and the definition
<luis-felipe>roptat: So, my inner boss allowed me to do some translation today. I'll start with the strings needing action that you pointed out.
<luis-felipe>roptat: I see in https://translate.fedoraproject.org/translate/guix/guix/es/?q=state%3A%3Ctranslated&offset=1 that the source string might have a typo.
<rekado>civodul: worked, thanks again!
<luis-felipe>roptat: Shouldn't it be "extension-graph" (singular)? Which is the name of a guix system action.
<permcu>Coming back to my earlier problem today: I learned the difference now. Fixed output derivations can access the network, normal ones not.
<permcu>In case anyone is interested too or can correct me maybe.
<roptat>luis-felipe, you can report typos in comments: click on comments below the string and select "Source string comment" as the scope, then comment in English
<roptat>I'll need to do a review of these comments at some point
<luis-felipe>roptat: roger
<apteryx>oops -- just reconfigured my system with the wrong config file
<sneek>Welcome back apteryx, you have 1 message!
<sneek>apteryx, civodul says: make-bootstrap.scm has all the tricks; one of them is to call remove-store-references to make sure that binaries actually work when run in the vacuum, such as an empty container
<apteryx>hehe, that old config.scm had the wrong Btrfs subvolume and a different way to specify the file systems. Had to wrestle a bit at the GRUB prompt, but got it to boot again. Now to reconfigure with the correct config ^^'
<rndd>hi guix!
<luis-felipe>Hi :)
<rndd>does anyone know, how to suspend guix laptop if you use exwm?
<rndd>or any other tiling wm
*luis-felipe does not know
<apteryx>seems rtl8812au-aircrack-ng is borked again
<apteryx>unbound variable: %build-inputs
<apteryx>jpoiret: I tried adding elogind to slim-service-type but still can't boot on core-updates-frozen, using slim + ratpoison
<apteryx>I'll try on a 2nd machine I have
<luis-felipe>What does "narinfo" stand for? And, is the format of this kind of file specified somewhere?
<jpoiret>apteryx: i hope it's not just a case of slim being way too old
<roptat>rndd, I think you cna do loginctl suspend or something similar
<apteryx_>jpoiret: I couldn't reproduce on a 2nd system (which boots much faster with less services and a SSD)
<apteryx_>so it's probably some service that delays another one, and causes the race/problem
<rndd>roptat: omg, it works, thank you =)
<rndd>roptat: but, do u know how to suspend and lock my screen?
<rndd>0_o
<roptat>no idea, sorry
<rndd>okay
<rndd>but thank you anyway
<apteryx_>jpoiret: I'll comment our my dockerd and nfs services and try again
<rndd>roptat: works for me "loginctl suspend && xlock"
<apteryx_>I suspect nfs, which seems to delay the boot by a long time may trigger the issue
<jpoiret>mhmmm, with the elogind dependency, i don't think there shoud be such a race issue
<jpoiret>did you check with herd status that xorg-server has the correct dependency?
<apteryx_>I just did now: Requires (user-processes host-name udev elogind).
<apteryx>jpoiret: slim is old but it works on an another machine and in VMs
<apteryx>it's just on that particular configuration/machine that it fails. I'll try with GDM for the kicks.
<Zambyte>Is there any way with guix to easily print out the definition of a package? Something like `guix edit` but it just prints the package to stdout instead of opening $EDITOR
<apteryx>
<cehteh>have you tried 'EDITOR=cat guix edit' ? prolly a terrible hack but may work
<roptat>that would print the whole file I think
<cehteh>yes
<Zambyte>It does print the whole file, and I can do things with head and tail to get it to what I need. Just wasn't sure if there was some command to do it in one go
<Zambyte>thanks anyways :)
<apteryx>jpoiret: I think it has to do with upowerd
<apteryx>it's the first service requesting 'org.freedesktop.login1'
<jpoiret>oh. Maybe your slim configuration asks for upowerd somewhere?
<jpoiret>ah, no, my bad, it's a desktop service
<jpoiret>well then, if it does indeed talk to elogind, it should have a dependency on it (which is not the case for now)
<apteryx>also upowerd:349: GLib-GIO-CRITICAL **: ... g_dbus_porxy_get_connection: assertion 'G_IS_DBUS_PROXY (proxy)' failed
<apteryx>(without the typos)
<tex_milan1>hello, I am using channels, have multiple of them. I cloned guix locally and want this local clone used instead of the guix channel. How to do that? I tried, but git pull says:
<tex_milan1>"Updating channel 'my-guix' from Git repository at '/home/milan/guix'...
<tex_milan1>guix pull: warning: pulled channel 'my-guix' from a mirror of https://git.savannah.gnu.org/git/guix.git, which might be stale"
<lilyp>apteryx: probably means you got a null pointer flying around
<apteryx>I think there's no dbus system session up at that point in time
<apteryx>later on I see also: Dec 7 14:45:57 localhost dbus-daemon[339]: [system] Connection has not authenticated soon enough, closing it (auth_timeout=30000ms, elapsed: 30005ms)
<apteryx>jpoiret: that's after adding elogind as a requirement to upowerd (didn't work :-()
<jpoiret>oho, true
<jpoiret>oh but hang on, that's a system service, right? i think you don't need an env var for the system bus
<roptat>tex_milan1, it's just a warning though
<roptat>tex_milan1, if you know what you're doing, you can ignore it
<tex_milan1>@roptat I am listening :) I want to modify existing service and I need to use my laptop as it is hardware connected so VM is out.
<jpoiret>apteryx: search results for that error all refer to fcitx, is it installed?
<nckx>Hi Guix.
<nckx>tex_milan1: The warning means that Guix is pulling from your local clone.
<tex_milan1>@roptat I understand it now :) Sometime I need time :)
<tex_milan1>thanks!
<roptat>:)
<nckx>It ‘knows’ (because the channel metadata tells it so) that it's not the real upstream, so it warns you that you're now responsible for keeping it up-to-date
<tex_milan1>i see now. originally I though it is telling me that it ignores my clone and uses the mirror at savannah. But it is the opposite what it really tells.
<apteryx>jpoiret: I wonder if it could be caused by clock adjustment
<apteryx>Dec 7 14:45:39 localhost shepherd[1]: Service dbus-system has been started.
<apteryx>Dec 7 14:45:48 localhost shepherd[1]: Service ntpd has been started.
<apteryx>Dec 7 14:45:24 localhost ntpd[344]: ntpd 4.2.8p15@1.3728-o Thu Jan 1 12:00:01 AM UTC 1970 (1): Starting
<apteryx>we just went 24 s back
<apteryx>that'd leave just 6 s for dbus-system to start without timing out, if it uses that system time
<apteryx>oh wait, it should give it 24s extra seconds?
<apteryx>jpoiret: not sure what fcitx is about, but since it works on another machine with similar config (slim + ratpoison), I guess it shouldn't be that?
<lilyp>apteryx: but what if dbus-system completes stating at say 14:45:25 and thinks something went wrong?
<lilyp>s/stating/starting/
<pyrotek45[m]>damn
<pyrotek45[m]>my emacs config doesnt work at all under guix
<pyrotek45[m]>rippppppppp
<jpoiret>does that ntp rollback happen on every boot?
<nckx>Timeouts shouldn't use system clock time.
<apteryx>yeah, they should use a monotonic clock... should
<apteryx>and if it's on Guile it doesn't support monotonic time at all, IIRC.
<apteryx>Guile's side
<apteryx>jpoiret: I've removed almost all my services, rebooting. We'll see
<apteryx>this should at least give me a cleaner, shareable /var/log/messages
<apteryx>also; should script something less awful than: "oh, you missed that password. press that reboot button again" in GRUB
<apteryx>s/reboot/reset/
<lilyp>pyrotek45[m]: look at the good side: Guix makes it a little harder to declare Emacs bankruptcy once you start using it :)
<guixy>Hi guix
<guixy>The default mpd user's home is in /var/run. It disappears on reboot
<guixy>It also has mode O700, so only the mpd user can make what it needs.
<guixy>or root
<apteryx>jpoiret: it seems to, yes!
<apteryx>this time it went back 25 s
<apteryx>I guess I could remove the ntpd service to see if it's caused by it
<pyrotek45[m]><lilyp> "pyrotek45: look at the good side..." <- what do you mean?
<apteryx>on a machine that boots fine the time is much more stable (time rewinds 1 s, not 25 s)
<KE0VVT>How do I run a command at a given time?
<apteryx>KE0VVT: you mean, like a cron job?
<tex_milan1>no success with [Controller] in bluetoothd config, that was a dead end. but if I run manually /gnu/store/xxx-blueman-2.2.3/libexec/blueman-mechanism then when I start blueman-manager I don't get that annoying error message. so either to start it in i3 config or find out why dbus doesn't start it :)
<KE0VVT>apteryx: Yeah, but I'm used to systemd timers.
<apteryx>you'll want to pass mcron jobs to the mcron-service-type
<apteryx>info '(guix) Scheduled Job Execution'
<apteryx>tex_milan1: you want your bluetooth controller to be enabled on boot, perhaps?
<tex_milan1>@apteryx ? yes, I autostart the bluetooth service, bluetooth works, but there is that annoying error message about Mechanism every time you start blueman-manager...
<apteryx>OK. Sorry, I don't know.
<KE0VVT>apteryx: thanks
<apteryx>jpoiret: bah, even without ntp-service-type the login failed
<gnucode>Hey guix!
<raghavgururajan>Any thoughts on this patch-set? https://debbugs.gnu.org/cgi/bugreport.cgi?bug=51885
<raghavgururajan>tex_milan | tex_milan1 : The link to blacksmith should now work from the above mentioned webpage.
<lilyp>pyrotek45[m]: with Guix, you're typically not installing packages from dozens of sources, so your init.el is slightly less prone to stuff breaking randomly
<lilyp>also rollbacks
<pyrotek45[m]>I I just install my packages using guix instead?
<tex_milan>@raghavgururajan thanks, will read it tomorrow
<pyrotek45[m]>s/I/So/, s/I/i/
<pyrotek45[m]>And they'll show up in emacs?
<jgart>Hi Guix, If I wanted to test out a new build system that is waiting review in upstream I can put it in a custom channel and dogfood it for a while?
<jgart>Hi lilyp, would you happen to have the time to review this one: https://issues.guix.gnu.org/52360
<jgart>It removes emacs-evil-magit, a deprecated package that has also been removed from melpa
<lilyp>We'd typically replace those by deprecated-packages first
<gnucode>jgart what new build system?
<jgart>lilyp, so I should add a deprecated alias to evil-collection?
<jgart>lilyp, evil-collection is what replaced emacs-evil-magit
<lilyp>that'd be my suggestion, but note that I don't use evil packages
<lilyp>so I don't know whether that'll work as drop-in replacement
<jgart>lilyp, well, the evil developers officially removed the package from melpa, released an official statement that you should now use evil-collection, and archived the evil-magit repo on github
<gnucode>raghavgururajan: that blacksmith thing loooks terrifying. I didn't realize that RAM was suffering from hardware bugs GNUtoo
<jgart>gnucode, php (composer-build-system), elixir, erlang, and the new python build system that follows pep517
<jgart>I'd like to put these build systems in a custom guix channel so I can try them out while I wait for them to get merged/finished
<gnucode>jgart: I'd love to know how you do that...I tried doing that for my opensmtpd-service with (guix records)...I'm currently testing it just without using a channel.
<apteryx>pyrotek45[m]: yes, it's best to manage all you Emacs packages with Guix
<gnucode>jgart: it would be cool to have a php build system!
<apteryx>gnucode: there's one by roptat nearly completed on the guix-patches tracker
<jgart>gnucode, https://git.sr.ht/~yoctocell/guix-yoctocell/tree/master/item/yoctocell/guix/build-system
<jgart>apteryx, gnucode yup that's the php one I'm talking about
<jgart>I'd like to put it in a channel and hack on it/dogfood it to see what's needed further before it gets merged
<jgart>ludo had suggested some fixes
<raghavgururajan>gnucode: Yep! Same here.
<jgart>apteryx, do you use evil-collection?
<apteryx>no
<jgart>raghavgururajan, would you like to hack on composer-build-system and packaging php packages in a channel?
<jgart>apteryx, oh ok, cool
<raghavgururajan>jgart: Hail GuixRus!
<raghavgururajan>jgart: I can't for now due to constraints you know off. :)
<jgart>ha yeah I can imagine
<raghavgururajan>s/off/of
<jgart>raghavgururajan, ok, I'll start putting them in the "Toy Store"
<drakonis>a toy store eh?
<drakonis>i'm a common lisp heathen now
<drakonis>i am empowered by my lisp knowledge
<jgart>drakonis, what's you workflow for using common lisp libraries with guix?
<drakonis>i'm currently running guix on a foreign system
<drakonis>not running it through guix at the moment
<jgart>drakonis, It's like Toys R Us but for software
<drakonis>but quicklisp is a gigantic boon
<raghavgururajan>drakonis: Are you from Canada by any chance?
<jgart>but it quicklisp compatible with guix?
<drakonis>am not
<drakonis>i believe it is, yes.
<jgart>how is quicklisp compatible with guix?
<raghavgururajan>Cool!
<jgart>I've been stumped on that
<drakonis>it runs on top of the common lisp implementation you chose
<drakonis>unless it has some kind of foreign dependency, there shouldnt be any issues getting it to run
<drakonis>the package you want to download with quicklisp
<drakonis>that is.
<jgart>but dont we want to develop a workflow for managing common lisp libraries with guix instead of quicklisp when using guix?
<drakonis>quicklisp stores all downloaded packages under quicklisp
<drakonis>definitely, yes.
<raghavgururajan>jgart: I just realized we have an outstanding package, that has to be transferred from GuixRus to Guix. hw-probe
<drakonis>oh, guix R us
<drakonis>hah
<jgart>hahah
<drakonis>where can i find that
<jgart>I'd like to make GuixRUs a community channel
<jgart>It needs some love
<raghavgururajan>jgart: We should make an announcement of GuixRus on guix-devel. ;-)
<drakonis>got an url for that?
<jgart> https://git.sr.ht/~whereiseveryone/guixrus
<jgart>Let's hack! GNU's joy store. It's worth the journey!
<drakonis>nice
<drakonis>this is a really good channel to have
<jgart>it's also here: http://src.whereis.みんな/
<jgart>running on a Guix System cgit service without certbot enabled
<jgart>I have to give that some love too
<drakonis>this might be the much needed community staging channel
<drakonis>guix-contrib
<jgart>The original idea was to make it a place where we can make pre-releases for patches in upstream
<drakonis>hmm, that could be good.
<jgart>but it's also a bazaar of packages
<jgart>another monorepo with various themes
<drakonis>hmm, it is missing podman
<jgart>software forks, custom packages, etc...
<drakonis>i've been fiddling with debian but will find time to write some packages
<drakonis>various interpreters and compilers that haven't been accepted into the main repository for reasons
<jgart>drakonis, If you'd like to add podman to it just send a patch to the canonical email for the sourcehut
<drakonis>i see
<jgart>I can review it and merge it
<drakonis>i believe i've seen it packaged by rekahsoft a while back
<drakonis>but it was on his repository
<jgart>If you'd like to have commit access we can talk about that also
<drakonis>we'll look into that later
<jgart>but maybe people find it easier to have me review their patches and do that work
<drakonis>do you want a channel under the guix irc org or separate?
<jgart>I'm less picky about what gets accepted than upstream. Commit messages don't have to follow GNU rules, etc...
<jgart>just do the right thing more or less
<jgart>ha
<jgart>We have an irc channel at #whereiseveryone on libera
<drakonis>i see
<drakonis>i'm already there
<jgart>drakonis, I sign all the commits for that channel. Raghav also has commit access currently
<KE0VVT>jgart: Wait, is there an issue with Podman?
<the_tubular>It doesn't exist in guix afaik ?
<KE0VVT>the_tubular: Is there an issue with it?
<the_tubular>I've tried my hand at writing a package definition a few weeks ago and I gave up. I think someone else was on the case, but I didn't follow closesly
<the_tubular>I'm really not experienced in go
<jgart>KE0VVT, not sure
<jgart>I've only used podman briefly
<jgart>KE0VVT, the_tubular drakonis https://issues.guix.gnu.org/52174
<drakonis>ah, nice.
<KE0VVT>jgart: I like Podman. I hope it has no FSDG issues.