IRC channel logs

<gabber>is it possible the official Guix 1.3 Image tries to install linux-libre-5.4.20-gnu.tar.xz -- which is now called linux-libre-5.4.20X.tar.xz? has that upstream link changed recently? https://linux-libre.fsfla.org/pub/linux-libre/releases/

<arescorpio>$ uname -r

<arescorpio>6.0.7-gnu ( linux-libre )

<apteryx>rekado: I had tried restarting nginx, and it seems shepherd got stuck; I can't 'herd status' anymore

<apteryx>and nginx is not serving pages

<apteryx>I guess we can reboot for the nth time

<user>ther some anothr herd runnign ?

<apteryx>no

<user>try to sudo strace herd status

<user>maybe filtering for open

<apteryx>yeah, I'm root

<user>it pproabbyl will try to open the unix socket to connect to shepherd

<apteryx>it's dead

<apteryx>strace shows it's stuck waiting for a read

<apteryx>I've rebooted berlin

<user>Oh :(

<user>I will sayd to check if connection if failling

<user>but if it trying to read th connection is ok

<apteryx>shepherd was upgraded today to resolve a memory leak; I hope it doesn't introduce anothing problem

<apteryx>*another

<apteryx>rekado: heya, apologies for the ping, but it seems the qla2xxx adapters are failing again :-/

<apteryx>ok, managed to boot berlin again but it's flaky

<apteryx>rekado: you'll have to do your screen dance for the issues<-debbugs rsync

<apteryx>pkill9: fyi, jami was just updated to 20221031.1308.130cc26 in 008f3a28c6

<crono>Quick question about package outputs in defining a package. I dont understand on how to specify what you want seperated from the regular output? I wanted to see if I could seperate the gnu hurd program from the documentation (in hopes to not install the whole program?).

<crono>Just for example, looking at the gettext-minimal definition, I dont see anything specifing the 'doc' part of what is commented as 9MiB of docs.

<lilyp>crono: depends on your build system, if you're e.g. using gnu-build-system the regular way, things should already be set up

<lilyp>if you have some weird makefile that you're fitting into it, you'll have to move things after install, see e.g. the ???move-doc??? phase in some packages

<ennoausberlin>Hello guix

<anschlussy>I think I've got COVID, so now I have an excuse to read the manual.

<ennoausberlin>issues.guix.gnu.org sis still down. So I want to report that bcl is broken on aarch64 again.

<ennoausberlin>*sbcl of course

<ennoausberlin>This happens way too often. Whoever updates the sbcl?? package definition please check if other architectures are building before pushing it. stumpwm and nyxt are dependent on it.

<unmatched-paren>morning guix!

<yarl>Hello!

<goober>hello!

<mekeor[m]>hello :)

<mekeor[m]>anschlussy: get well soon!

<mekeor[m]>channel-with-substitutes-available should accept multiple servers and also should be able to specifically check if there substitutes for the packages installed on the system

<anschlussy>thanks, mekeor[m]

<rekado>apteryx: I don???t see anything indicating a hardware problem in the hardware logs.

<rekado>apteryx: we should aim to get the websites onto bayfront so we can reboot the machine at will

<rekado>or onto 129

<rekado>if there???s a hardware problem Dell will request diagnostics, and collecting the diagnostics report might lead to reboots

<cbaines>bayfront is already building the website, and has been doing for a while

<rekado>I mean: actually flipping the switch to serve the sites from there

<cbaines>indeed, I think some of the needed work was done, but maybe not all

<nckx>...after I do that and refresh for the eleventieth time, it responds. Still, looks like I chose the perfect weekend for a relaxing holiday.

<nckx>Still QLA nonsense?

<yarl>I wonder how to clean /run/current-system/profile/lib/guile/3.0/site-ccache/guix/status.go as it is a read-only filesystem.

<yarl>s\guix/status.go\\

<yarl>I am playing with geiser and it keeps complaining about source file newer than compiled.

<minima>(not very important: i had the impression that org was included in emacs by default, but i see there's actually a separate guix package for it, emacs-guix?)

<minima>hrm... sorry... emacs-org

<cbaines>minima, the version included in emacs and the one packaged separately are different

<cbaines>I think the one included in emacs is generally older

<minima>cbaines ah, understood, thanks!

<lizog>i'm trying to pack a python-written software, but using gnu-build-system(they use Makefile to install instead of setup.py). how am i going to make the python scripts properly find their dependent packages?

<lizog>ok i got the mistake... when doing wrap-program i typed `GUIX_PYTHONPATH' as `GUIX_PYTHON_PATH'

<lizog>which file should i place if i'm packing https://syncplay.pl/

<awb999>which speed do you guys get when downloading substitutes?

<awb999>I am getting 100k-250k download speeds,

<awb999>which means every build that needs substitutes needs hours to download.

<awb999>this makes it really hard to make the smallest changes to my config,

<awb999>as I have to wait hours after a small change...

<awb999>and it happens EVERY DAY

<awb999>since guix changes so often...

<rekado>awb999: depends on your ISP

<rekado>the server has a good connection, as does the data centre, but it???s downhill from there

<rekado>if you want good download speed from everywhere in the world you need distributed caches

<rekado>it???s not something I???ll set up

<awb99_>Distributed caches?

<awb99_>How could that work?

<rekado>CDN

<awb99_>It would help if substitutes would be shared at leastnon the same machine

<awb99_>Or the local network

<awb99_>Guix on one machine needs to download

<awb99_>1. For the system

<awb99_>2. For the user

<awb99_>3. For guix home

<singpolyma>Right now do we even know if substitutes can be served by a static http server?

<awb99_>So if the machine would share them .. this would help

<awb99_>Also multiple maxhines in a network all will need mostly the same substitutes.

<rekado>awb99_: this literally already exists

<awb99_>Which download speed do you guys get?

<awb99_>Normally?

<singpolyma>It's never been slow enough for me to notice

<singpolyma>But I also only guix pull every few months

<awb99_>Wow

<rekado>awb99_: check Invoking ???guix publish??? in the manual, specifically ???--discover???

<rekado>sorry, ???--advertise???

<awb99_>Great! Thanks rekado

<apteryx>awb99_: I get 2-3 MiB/s download speeds from the other side of the Atlantic

<awb999>thanks!

<awb999>good to know!

<apteryx>it doesn't feel like the bottle neck

<awb999>must be a bad international internet to south america.

<apteryx>rekado: I see. Hmm. else perhaps some regression in Linux 6.x in qla2xxx, that somehow corrupts the adapter states for a couple reboots? that's far fetch.

<VesselWave>Hey Everyone! I am trying to share one guix system config across multiple machines, base-system: paste.debian.net/1260461 sigma: paste.debian.net/1260463 . But when I try to reconfigure `sudo -E guix system -L ~/dotfiles/.config/guix/systems reconfigure ~/dotfiles/.config/guix/systems/sigma.scm` I get an error `error: base-operating-system: unbound variable hint: Did you forget a `use-modules' form?`. What could I do?

<VesselWave>Correct links. base-system: paste.debian.net/1260467 sigma: paste.debian.net/1260466

<PotentialUser-30>i did "guix install nheko" och want video call - got error message : Missing gstreamer plugins: qmlgl

<PotentialUser-30>i think it's in gst-plugins-good and we need compile with QT5 - so what can i do to finally get the libgstmlgl.so ??

<PotentialUser-30>I Meant: libgstqmlgl.so

<ae_chep>I have a question: I want to edit the commit msg for a ticket I already submitted. Normally I would do ammend an existing commit. What do I do in this case. I think I do something with "reroll-count" but I haven't fully understood yet

<gnucode>PotentialUser-30: sounds like you need to dig into updating package definitions. Best of luck. I've never done that.

<apteryx>PotentialUser-30: perhaps we can have a gst-plugins-good-qt variant that pulls in qt

<apteryx>it's used in gtk@4 so it's undesirable to have qt in its closure

<PotentialUser-30>File a bug against nheko? Against gst-plugins-good? How is i usually done? Craft a VARIATION ??

<apteryx>yes, a variant package (define gst-plugins-good-with-qt (package (inherit gst-plugins-good) ...))

<PotentialUser-30>Afterwards install gst-plugins-good-with-qt myself or also a variant for nheko that depends on it ?

<apteryx>nheko could use it as inputs instead of gst-plugins-good

<apteryx>assuming nheko is some leaf packgae

<PotentialUser-30>Thanks - i'll try?? - but first i think i have to read a lot...

<formbi>hi

<formbi>how to make a guile module available to an mcron job in the system config?

<formbi> http://ix.io/4fII

<formbi>I'd like to use srfi-19 here

<formbi>but it says that the variables are unbound

<apteryx>could someone please restart https://ci.guix.gnu.org/build/1692561/details ?

<apteryx>I'll recreate my credentials later

<mothacehe>apteryx: done

<gnucode>formbi: that sounds like a fun project. What are you trying to do ?

<mothacehe>I sent a mail about CI credentials on guix-sysadmin

<formbi>gnucode: I'm trying to make a job which creates a btrfs snapshot and names it with a timestamp

<apteryx>mothacehe: thanks!

<gnucode>formbi: woah! you don't start out small do you? hahah

<apteryx>eew, gst-plugins-bad depend on python2 through fluidsynth and lash

<formbi>I think it shouldn't be hard (see the link), but the jobs don't seem to have access to guile modules

<apteryx>mothacehe: they simply expired, right?

<apteryx>formbi: I believe there's an example about how to do that in the manual

<mothacehe> yes but we also lost the machinery to create them in the berlin SAN transition

<apteryx>ah! oops

<apteryx>perhaps we should commit that machinery to the maintenance repo

<formbi>apteryx: in mcron or guix?

<apteryx>formbi: for an mcron job when using it with guix system

<gnucode>formbi : info guix RET i mcron RET

<VesselWave>mekeor[m]: You aswered me about org mode issue. My config is based on daviwil's: https://github.com/daviwil/dotfiles/blob/master/Systems.org. Mine is work in progress, it even can't be used to reconfigure :)

<gnucode>you may need to add in a (add-to-load-path (dirname (current-filename)))

<formbi>apteryx: ah, I see it

<mirai>What's the cause for "guix system: error: primary group 'x' of user '#<<location> file: myconfig.scm' is undeclared"?

<mirai>I'm using define-configuration and have default-values defined so it shouldn't be the case

<formbi> http://ix.io/4fIW that should do the job, right?

<VesselWave>mekeor[m]: Try to download his config and put cursor after '>' symbol in %my-desktop-services declaration. It will be hightlighted, but it shouldn't.

<gnucode>I'm starting to realize that as I code a service in define-record-type* and try to make it super awesome... I am slowly (without realizing it) poorly rewritting (define-configuration).

<mirai>Do I need to use sudo with './pre-inst-env guix system reconfigure' when writing a new service definition?

<apteryx>to try them out with your system? yes

<apteryx>you'll also need to prefix with sudo -E

<apteryx>I'd recommend writing a test first, though

<euandreh>mirai: yes, because it'll change root owned things in /var/guix/

<apteryx>else you could end up with a stuck shepherd and an unusable system

<mirai>apteryx: test?

<apteryx>(you can always boot to the previous gen though)

<apteryx>mirai: system test

<apteryx>'make check-system TESTS=jami' for example

<euandreh>or launch a container

<mirai>do I add "sudo" to "guix environment guix --pure --ad-hoc ..." ?

<euandreh>no need

<euandreh>even a `./pre-inst-env guix system build ...` works without sudo, because up untill then you're only adding data to the store

<mirai>euandreh: I get an error message of "guix system: error: primary group 'x' of user '#<<location> file: myconfig.scm' is undeclared"

<mirai>which I am not sure if its due to the lack of sudo

<euandreh>mirai: when running `guix environment guix`?

<mirai>apteryx: It doesn't seem to do anything https://paste.centos.org/view/f7306577

<mirai>mirai: yes

<mirai>euandreh: yes

<euandreh>hmm, I think the problem is something else. Are the permissions on "myconfig.scm" 644?

<mirai>yes

<euandreh>mirai: oh, but myconfig.scm is at the end of the `guix environment guix` command. What is the full line that is giving you an error?

<a12l>Is there a way to generate a default config? Testing Guix out as a package manager on NixOS.

<mirai>euandreh: https://paste.centos.org/view/4c194682

<euandreh>mirai: that will need a sudo

<euandreh>probably sudo -E

<euandreh>a12l: with guix as a package manager, there is no configuration file to generate

<euandreh>you can use Guix Home, and have a configuration for that

<euandreh>or maybe a file with a list of packages

<euandreh>a12l: but the nixos-generate-config equivalent is for the full OS, the Guix System

<euandreh>a12l: when installing Guix System, the TUI installer generates an initial configuration for you

<a12l>euandreh: Thanks!

<euandreh>but there are examples for the OS config in the Guix Cookbook, and examples of the home environment in the manual

<euandreh>cookbook: https://guix.gnu.org/en/cookbook/en/guix-cookbook.html

<euandreh>manual: https://guix.gnu.org/manual/en/guix.html

<mirai>euandreh: what was the container approach you were saying before

<euandreh>a12l: np

<mirai>is there a easier way to test changes that can potentially break the system?

<euandreh>mirai: just swap "reconfigure" for "container"

<euandreh>`./pre-inst-env guix system container ~/xfer-guix/systems/test-nuc.scm`

<euandreh>It drops you in a container with that configuration

<euandreh>I'd say this is the simplest one

<mirai>the pre-inst-env script is executed within the "guix environment" or outside of it?

<mirai> https://guix.gnu.org/manual/en/html_node/Building-from-Git.html

<mirai>I'm getting the same error with "system container" (running from within the guix environment)

<unmatched-paren>evening guix!

<gnucode>unmatched-paren: what up!

<tex_milan>Hi, how do you debug scm? I have my package and I know that one package is causing the issue, but on first sight there is nothing I see wrong. Error I get is just about unbound variable.

<unmatched-paren>tex_milan: it's hard to debug guile

<unmatched-paren>unfortunate flaw

<tex_milan>I noticed! I hoped there is some trick :D Or validator.

<unmatched-paren>nope, sorry :)

<unmatched-paren>what's the full error? (paste.debian.net it pls)

<tex_milan>in procedure proc: unbound variable NAME_OF_MY_PACKAGE

<unmatched-paren>could you paste the package file?

<tex_milan>did you forget a `use-modules' form?

<tex_milan> https://paste.debian.net/1260478/

<unmatched-paren>whole file, please

<tex_milan> https://paste.debian.net/1260479/

<tex_milan>every other package in that file works, when I comment out the python-tflite-runtime

<tex_milan>it is possible to install tflite-runtime with pip3 and it works

<euandreh>mirai: back again. Looking at the latest pastebin link you sent, the error wasn't permission related

<euandreh>but of a missing group

<awb99_>When I build a guix image I would like to include some directories at build time. This directories should be say in /var/app1. A lot of times I need to add something that is not managed by guix.

<awb99_>Say I am writing annimage tona sd card

<awb99_>It would be really helpful to automatically add some files to it.

<mirai_>euandreh: do I need to explicitly add the group to operating-system ?

<mirai_>does the service not automatically create it?

<unmatched-paren>tex_milan: hmm... i'm not sure what's happening :(

<euandreh>mirai: you shouldn't have to add the group maually. I guess the service is creating it by extending "account-service-type"?

<mirai>euandreh: https://paste.centos.org/view/557acc10

<mirai>it already is extending it

<mirai>unless I'm doing something really wrong here

<euandreh>it looks really fine

<euandreh>mirai: does it get enabled correctly without you adding the group to any user?

<mirai>it doesn't get past system container or reconfigure

<bdju>anyone else using ranger and getting this error when moving over files? Loader work process failed: Getting preview of ... (file name here)

<mirai>euandreh: https://paste.centos.org/view/53d35130

<mirai>I am running all of this from within the guix environment for guix development as described at https://guix.gnu.org/manual/en/html_node/Building-from-Git.html

<euandreh>mirai: does it give you the same error if you inline the string "mympd" in line 130, 132 and 133?

<euandreh>lines referring to https://paste.centos.org/view/557acc10

<mirai>it's starting to fetch substitutes after inlining

<mirai>I presume it's starting to work?

<euandreh>I guess so

<mirai>?? why does inlining fix it?

<mirai>there are defaults already defined though?

<euandreh>mirai: were you using the default values when enabling the service?

<mirai>I used https://paste.centos.org/view/efe6633b

<euandreh>mirai: well, the only funny thing that led me to try inling is that I think you want to change line 130 from "(name user)" to "(name group)"

<tex_milan>#unmatched-paren any idea how to debug that?

<unmatched-paren>tex_milan: there's nothing that immediately comes to mind... :(

<euandreh>mirai: otherwise having values for user/group that aren't identical would break that config

<mirai>euandreh: even though the strings for both user and group are the same?

<mirai>I am not too sure of that code segment myself, I took it from https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/services/version-control.scm#n311

<euandreh>well, I was trying to guess that you had different values for that, and that was the reason it was breaking

<euandreh>mirai: I think that you just found a bug :)

<mirai>euandreh: bug in gitolite or in guix?

<euandreh>gitolite

<mirai>meanwhile the container command has finished, what's "next"?

<euandreh>for the case where one doesn't use the default value for the config, and choose different strings for user and group

<euandreh>mirai: well, it drops you into the container, no?

<mirai>eeeh, I have no idea

<mirai>it still has the same "prompt" as before

<euandreh>mirai: if you type exit, does it close the shel?

<mirai>it quit the environment

<civodul>hi! does anyone experience the bug whereby "herd restart nginx" causes shepherd to block? https://issues.guix.gnu.org/58926#6

<civodul>i'm looking for additional debugging info

<euandreh>civodul: I had this problem once, when I was interacting with the shepherd on the activation script of a service. Things like "(restart-service ...)"

<euandreh>mirai: see, I know it was just staring at us

<euandreh>mirai: if you run the container command, it spits a bunch of build data to STDERR, and a single line to STDOUT

<euandreh>mirai: that line is the script to run, that drops you into a container

<euandreh>mirai: so you can just run that by puttin the command in backticks

<euandreh>$ `guix system container ...`

<mirai>ah, right

<mirai>it was obscured by a warning about remaining system space

<euandreh>yes, a by a bunch of build logs

<mirai>In procedure fport_write: Operation not permitted

<mirai>does this require sudo?

<euandreh>IIRC, i think it does

<mirai>euandreh: if you want, you can open the bug report for gitolite as I'm not too sure about how it's supposed to work

<euandreh>mirai: ACK, I'll send a patch and Cc you

<mirai>euandreh: thanks

<euandreh>civodul: I'll try to recreate it in a vm, and I'll send it to you if I do

<euandreh>mirai: np :)

<mirai>how do I ignore the exception thrown by delete-file-recursively when the directory doesn't exist?

<unmatched-paren>mirai: why not (when (file-exists? "dir") (delete-file-recursively "dir"))?

<mirai>unmatched-paren: that can work too

<lop>Hi, I'm having an issue with Gnome not working when I login whenever I pull and then reconfigure

<unmatched-paren>lop: you have any gnome extensions?

<lop>I didn't install any myself, how would I check for them?

<unmatched-paren>oh, if you didn't install any you don't have any

<lop>I had this Gnome issue about two months ago, my solution since then has been to just not pull and update my whole system

<euandreh>mirai: I just realized I don't have you email address XD

<euandreh>mirai: I just sent the patch, let me grab the link

<unmatched-paren>lop: could you paste.debian.net your system config and guix package --list-installed?

<mirai>euandreh: you need my email?

<lop>I'm in a Sway session right now and I don't know what's going on with the clipboard but I'll try to figure something out

<unmatched-paren>clipboard issues with sway seem to be common here but i've never seen any myself...

<unmatched-paren>maybe it's because i'm using sway 1.7 instead of guix's 1.6

<euandreh>mirai: not anymore XD. Its because I said I'd CC you in the gitolite bug, but only later realized that we were discussing this on IRC

<tex_milan>unmatched-paren: maybe it was due to some whitespace? I copied working package and replaced it with nonworking line by line and guile doesn't complain now. well, pypi-url fails to fetch it... can't find the package on pypi although that package is there and I can install it with pip3 install tflite-runtime...

<unmatched-paren>tex_milan: odd

<unmatched-paren> https://pypi.org/search/?q=tflite-runtime <- i can find it on pypi

<mirai>euandreh: I think I know what's going on with inline/no-inline

<euandreh>what is it?

<mirai>I am using match-lambda for a record but I am not passing all the fields it has

<euandreh>mirai: hmm, that makes sense. Does a (display ...) confirms that?

<mirai>I used the repl

<mirai>it has user: #f and so on

<mirai>so I'm guessing that's it

<tex_milan>yes, it is there, but method url-fetch with uri pypi-uri "tflite-runtime" "2.10.0" can't

<lop>Oh it was just nvim doing something weird

<lop> https://paste.debian.net/1260489/

<mirai>I thought match for records would just ignore the missing fields

<tex_milan>could not find its Disarchive specification

<mirai>and pick the first N fields but the repl seems to suggest that's not the case

<lop>This is a pretty atrocious config but the times I finally got things working I usually didn't have any patience left to clean up things

<unmatched-paren>lop: hmm, that looks normal...

<unmatched-paren>lop: could you poke around it /var/log/gdm, perhaps?

<mirai>is there a match-lambda equivalent but only for a subset of a records fields?

<tex_milan>oh, maybe it doesn't have source package, only pre-builded one, that can be installed with pip3

<unmatched-paren>tex_milan: i guess you could switch to git-fetch

<tex_milan>python-build-system will try to compile it? not only install binary package?

<unmatched-paren>tex_milan: ?

<euandreh>mirai: dunno, it sounds reasonable that one would exist

<euandreh>mirai: it indeed doesn't ignore the first n fields. I knew that and checked the code, but missed it

<euandreh>I thought it was being done correctlu

<euandreh>mirai: here is it https://issues.guix.gnu.org/59223

<euandreh>the hardest part is writing a good commit message :)

<mirai>euandreh: neat, thanks

<tex_milan>unmatched-paren: guix import: error: no source release for pypi package tflite-runtime 2.10.0, his indicates that the package is available on PyPI, but only as a "wheel" containing binaries, not source. To build it from source, refer to the upstream repository at `https://www.tensorflow.org/lite/'.

<tex_milan>unmatched-paren: I don't really want to build it from source. it is pain even on normal system I heard

<unmatched-paren>tex_milan: if you use the upstream repo with git-fetch, it will build it, yeah

<unmatched-paren>oh? how so?

<lop>unmatched-paren: I'm not sure exactly what to look for, but there's a few greeter files and the last one has "gnome-session-binary[7748]: WARNING: Lost name on bus: org.gnome.SessionManager" before a bunch of connection closed and unloading messages

<mirai>euandreh: how do I later delete the container?

<euandreh>mirai: just "exit" it

<unmatched-paren>lop: when does this happen, specifically? as soon as you log in?

<mirai>euandreh: the script in the store gets automatically gc'd?

<euandreh>not automatically, but eventually

<euandreh>eventually, when you run "guix gc"

<euandreh>not only the script, but also everything it depends on and that it did build to create the script

<lop>unmatched-paren: Yes, once I enter the correct password it just kicks me to an "Oh no! Something went wrong" white screen

<tex_milan>is there some python binary only package in guix? I guess no, right?

<unmatched-paren>tex_milan: no

<tex_milan>ok, I write a note to package description to ask user (me) to install it manually with pip3, that works. thanks for talk!

<tex_milan>is it possible to execute a bash commands in home-manager during reconfiguration? I would put it there.

<unmatched-paren>lop: maybe try ``dconf write /org/gnome/shell/disable-user-extensions true''?

<unmatched-paren>just in case there is somehow an extension in there...

<mirai>euandreh: how can I test a network service (in this case mympd) listening on port 80 within the container?

<mirai>I've got it running and opened a shell to it

<lop>unmatched-paren: I'm getting an error saying I can't autolaunch DBUS without X11 $DISPLAY

<unmatched-paren>lop: maybe you could try using gnome wayland...?

<unmatched-paren>you need to enable gdm-service-type's (wayland? ...)

<unmatched-paren>and select gnome (wayland) in the session menu

<euandreh>mirai: IIRC, you just need to enable networking in the container

<euandreh>use the -N flag

<euandreh>I think it is "guix system -N container ..."

<euandreh>mirai: "-N" or "--network"

<euandreh>yep, that's it, and now a) your container can access the network, and b) you can access the container via the network

<mirai>euandreh: interestingly I get different results for an "equivalent" command

<mirai> https://paste.centos.org/view/d830f118

<mirai>I can't access port 80 from the host side :(

<civodul>mirai: FWIW i've also been unable to use nsenter and "guix container exec" lately

<civodul>looks like something broke

<mirai>in fact, not from within the container too

<civodul>worth a bug report

<mirai>-bash-5.1# herd status

<mirai>error: connect: /var/run/shepherd/socket: No such file or directory

<mirai>civodul: I think my guix is still from a few days ago

<mirai>civodul: guix describe: guix c52cdd1, commit: c52cdd18d6f8dddd69dfb1f8b5214a0db6ebb562 (Nov 08 2022)

<mirai>nsenter still functional

<mirai>(?)

<lop>unmatched-paren: Honestly I'm feeling up to just stick with Sway at this point

<lop>I will try switching to sddm and using gnome-wayland if I do need to use it for some reason

<lop>Thank you for helping me try to debug the monstrous labyrinth that is Gnome

<unmatched-paren>lop: Fair. If you want a better login manager, try using greetd-wlgreet-sway-session ;)

<Wurt>unmatched-paren: do you have a working configuration using that login manager? In my laptop the vt terminal using it only shows a blinking cursor.

<rekado>apteryx: a more mundane explanation could be the combination of actual hardware failure on the SAN and our lack of multipath.

<unmatched-paren>Wurt: https://git.sr.ht/~unmatched-paren/conf/tree/root/item/system.scm#L77

<unmatched-paren>Wurt: oh, yeah, you need those greeter-supplementary-groups, you might be missing them

<rekado>apteryx: there is a broken disk in the SAN; by itself that???s not any reason for concern. But there???s ongoing work to replace it and run diagnostics.

<unmatched-paren>although seat isn't needed if you're on elogind

<rekado>this *might* have an impact on systems that are not redundantly connected to the SAN.

<Wurt>unmatched-paren: Thanks, I will try it.

<civodul>apteryx: reading the log, IIRC you rebooted due to shepherd being stuck after "herd restart nginx" or similar

<civodul>there's a known bug in this area that i'm looking at: https://issues.guix.gnu.org/58926

<civodul>it non-fatal though: to work around it, "strace -p1" to see the wait4 call, and then kill the PID in there

<civodul>(terrible, i know)

<podiki[m]>civodul: are there any plans for any freezing or big merges for 1.4.0? or just roughly set a tag (for translations?) whenever the last blocking bugs are fixed?

<civodul>podiki[m]: we'll create a branch when we're done with the major bug fixes

<civodul>and there we'll only take important fixes

<civodul>and translation updates

<civodul>(should be a short-lived branch)

<sharlatan>Hi folks, is disagrchive has TLS expired?

<sharlatan>>> guix lint: warning: failed to access Disarchive database at https://disarchive.ngyro.com/: TLS certificate error: X.509 certificate of 'disarchive.ngyro.com' could not be verified:

<podiki[m]>civodul: gotcha. then after that I guess is a core-updates cycle? :)

<podiki[m]>sharlatan: seems that way, shows as expired on 11/11

<vagrantc>have to make sure the stable release is outdated, of course! :)

<sharlatan>By the way I've fix some issue with python-asdf build after recent staging->master merge =), I guess it's the last set of patches fixing it https://issues.guix.gnu.org/59113

<civodul>podiki[m]: core-updates merge even, like vagrantc pointed out :-)

<sharlatan>civodul: when it's planned by the way? core->master

<civodul>sharlatan: unfortunately that server is not under the control of the Guix Dream Team

<civodul>sorry, Sysadmin Team

<civodul>sharlatan: it's not planned :-)

<civodul>first trying to get that release out i guess

<podiki[m]>very good, I love the excitement of all the newness in core-updates

<civodul>yup!

<mirai>Is sheperd available within a guix container or do I have to use 'vm' for it?

<podiki[m]>the release will be nice to summarize all the cool work since 1.3.0 (and hopefully be clear to newcomers what we mean by "release", I found that confusing at first)

<sharlatan>May someone have a look at #59113, I'm keen to continue with other Astrop patches in my queue ^.^ https://git.sr.ht/~hellseher/ffab/tree/main/item/ffab/packages/astronomy.scm

<vagrantc>a release for guix seems to be an excuse to get out a fresher installer and also a bootstrapping checkpoint so people don't have to start from ancient guix

<unmatched-paren>Perhaps we could use 1.4.0 as an opportunity to move /manual -> /manual/1.4.0 and /manual/devel to /manual? :)

<vagrantc>and maybe something downstreams can package :)

<vagrantc>ACTION whistles innocently

<civodul>:-)

<podiki[m]>unmatched-paren: agreed; I was thinking have just one manual link, with a note on the first page (and maybe in installing section) that the 1.4.0 snapshot exists for use with initial installation if using the 1.4.0 image

<vagrantc>maybe /manual/current and /manual/VERSION

<vagrantc>or yeah, just the current manual

<podiki[m]>maybe a banner at the top of 1.4.0 (or whatever version) noting it is not meant for current usage after install/guix pull the first time

<vagrantc>the utility of the non-current manual is debateable to anyone other than archaeologists

<sharlatan>or more commont /manual/lates /manual/<version-tag>

<sharlatan>r

<podiki[m]>install is the only reason I can think of

<podiki[m]>though there is info )\

<podiki[m]>(sorry cat)

<podiki[m]>though there is the info manual in the installation environment, can be easier to browser on the web on another device

<vagrantc>wonder if guix pull shouldn't list news items in reverse order ... just got one where the top line was "Linux-libre kernel updated to 6.0" and the bottom was "Linux-libre kernel updated to 5.19" ... almost makes it look like it was downgraded

<vagrantc>not hard to figure out ... but

<podiki[m]>I think reading between the lines, it is tell you to guix pull more often :) (but point taken on say the first pull after being away a while or install)

<vagrantc>with this machine, i pulled about as often as i booted it, can't pull more often than that!

<mirai>is 'herd' not available at all with 'guix system container'? I'd like to test this service definition without breaking the main installation

<pkill9>dunno, but there is guix system vm which should run shepherd

IRC channel logs

2022-11-12.log