IRC channel logs

2021-08-16.log

back to list of logs

<slyfox>is there a verbose way to trace 'guix refresh' operation? I wonder what would be the easies way to debug failure: '$ ./pre-inst-env guix refresh mc' -> 'gnu/packages/mc.scm:41:13: warning: 4.8.26 is greater than the latest known version of mc (4.6.1)'
<str1ngs>slyfox: I think that output is intended. but it's possible your guix is older? what verion of mc does guix show mc produce?
<slyfox>It's: 8d80a708535546394deaea6489245ecc09b29131 'Date: Sun Aug 15 16:58:40 2021 -0400'
<slyfox>should not be too old
<slyfox>note that refresher reports a severe downgrade: 4.8.26 -> 4.6.1
<str1ngs>ahh so the refresher is not parsing versions correctly then?
<slyfox>i guess it incorrectly parses html (or something like that) and i wonder how i could peek at it without sprinkling debug statements
<str1ngs>how do I display of off the outputs that rely on an input?
<str1ngs>all of*
<ruffni>str1ngs: i think `guix gc --references`
<ruffni>wait. do you mean outputs of one package or all outputs in guix?
<str1ngs>I just need to see what outputs use g-golf an input. I know it can be down. just forget the snippet :)
<str1ngs>as an*
<str1ngs>done*
*str1ngs notices typing has gone to crap :(
<lfam>Outputs? Or packages?
<lfam>Outputs don't really have inputs, but packages do
<str1ngs>output's
<str1ngs>less use package for sake of simplicity :)
<lfam>Packages have inputs and produce outputs. Outputs retain references
<lfam>It would help if you just explained what you are trying to do
<lfam>If you want to know which packages depend on g-golf, you can use recsel
<str1ngs>I have explained already
<lfam>`guix package -s . | recsel -e 'dependencies ~ "g-golf"' -p name,synopsis`
<lfam>You asked a question but it wasn't clear
<lfam>There's also `guix refresh --list-depenendents` which will tell you what will be rebuilt if you change g-golf
<lfam>If you want to check on actual build outputs, you need to build everything and then query the database of references
<lfam>* `guix refresh --list-dependents`
<str1ngs> guix refresh -l g-golf works thanks
<roptat>I'm trying to have clang generate makefile dependencies, but it complains one of the header files is missing. Indeed, I'm trying to make it generate that so that make knows it has to generate that header first... is there a way to solve this?
<iskarian>sneek, later tell yoctocell: https://issues.guix.gnu.org/50072 :)
<sneek>Will do.
<iskarian>sneek: botsnack
<sneek>:)
<lfam>roptat: Not sure I follow :)
<roptat>I'm building a cpp file that includes a generated header. There's no makefile, it's the soong build system for android, so I generate one from the Android.bp description. The makefile needs to generate the header before it attempts to build the cpp file
<roptat>so I thought I could generate a dependency file with the C compiler (-M and friends), but that seems to require all the headers already
<roptat>(I need something like "a.o: b.h", and there are makefile rules for building b.h and a.o)
<lfam>Wow
<muradm>i wan't to create /tmp/.X11-unix/ directory on boot after /tmp is mounted, what would be the best way to do that?
<muradm>any service like etc-service-type to make directory?
<muradm>one way to make it is probably one shot shepherd service
<muradm>any better way?
<roptat>oh no, I thought I was done with adb, but I'm still missing two dependencies I hadn't seen...
<roptat>ok, more work: some java stuff for some reason...
<str1ngs>DigitalKiwi no more archlinux for you?
<str1ngs>hello raghavgururajan is there an existing branch for gtk4?
<raghavgururajan>str1ngs: There is no separate branch for gtk4, but wip-gnome branch is used for gtk4.
<raghavgururajan>There updates in that branch that is required by gtk4.
<raghavgururajan>The patches in #48554 are to applied and developed on wip-gnome branch.
<raghavgururajan>sneek, later tell char: I have replied to your message in #48554.
<sneek>Okay.
<the_tubular>Damn, everytime I get something to work on guix it impress me. I really like that package manager
***roptat_ is now known as Guest5707
***Exagone313 is now known as Exa
<abrenon>good morning
<thorwil>hi! geeqie used to work, but now fails to start with “(geeqie:15752): Gdk-ERROR **: 09:18:20.939: The program 'geeqie' received an X Window System error.” and “Details: serial 176 error_code 11 request_code 149 (unknown) minor_code 2”
<thorwil>this is on a foreign distro, so i would like to know, if anyone can reproduce that.
<thorwil>i found https://www.pclinuxos.com/forum/index.php?topic=154307.0 which says `qeeqie --disable-clutter` should help, but it doesn’t, here. that option also doesn’t appear in the man page.
<PurpleSym>thorwil: Works for me on a foreign distro. `guix pull` is from this morning.
<efraim>it also works for me, enlightenment on guix system. I also have problems with gtk apps more often than some others
<thorwil>thanks for checking!
<yoctocell>sneek: later tell iskarian: I saw your patch for the `git-fetch' updater, and it worked great when I tested it on a few packages hosted on GitHub.. Thank you for the work! :-)
<sneek>Welcome back yoctocell, you have 1 message!
<sneek>yoctocell, iskarian says: https://issues.guix.gnu.org/50072 :)
<sneek>Will do.
<yoctocell>sneek: later tell iskarian: I think the next step would be to add a `generic-git' updater, so that it works on any Git repo, and not just ones on GitHubl
<sneek>Okay.
<DigitalKiwi>str1ngs: https://twitter.com/ArchKiwi/status/1252329061902401538?s=20
<DigitalKiwi>i have an assortment still lol
<cehteh>where do i setup routing in guix esp routing through wireguard?
<bricewge> cehteh Did you read the cookbook about wireguard?
<cehteh>yes
<cehteh>which is pretty sparse
<vldn>ArneBab anyway to convert normal guile scripts to wisp?
<cehteh>bricewge: i have wireguard on my router and on a bastion-host and want to connect from the outside to my bastion host, then have routes into my internal networks
<cehteh>the internal hosts/networks are not connected to wireguard
<cehteh>i have wireguard running, but dont know how to set up routes
<cehteh>also ok to route *all* traffic through wg
<bricewge>Ok, then your question is more wireguard related than guix
<cehteh>no
<cehteh>i dont know how/where i setup routes in guix especially
<cehteh>in debian i can add pre-up hooks and stuff
<bricewge>It depend on the networking service you are using
<cehteh>guix (service wireguard-service-type
<bricewge>No, the shepherd service providing `networking` https://guix.gnu.org/en/manual/devel/en/guix.html#Networking-Services
<cehteh>as in network manager? happend to be by default and its ok because its a laptop (sometimes users have to reconfigure wlan/lan)
<bricewge>Like `simple-networking-service-type`, `network-manager-service-type`, etc
<bricewge>Like `simple-networking-service-type`, `network-manager-service-type`, etc
<cehteh>but in network manager wireguard cant be added through the ui, i didnt bother to go try with nmcli to add wg and added a wireguard service to the system config
<cehteh>ah in network manager i still see the wg0, guess i can add routes from there, but sounds a bit odd to have configuration scattered over very different places
<bricewge>If I were you I would use networkmanager entirely, and configure the route in it
<bricewge>ie. not using `wireguard-service-type`
<cehteh>i tried that but 'add an vpn' didnt shown any modules
<bricewge>It's odd you can't use nm-applet to import a wireguard config
<cehteh>ah
<cehteh>lemme try it in another way
<bricewge>Okay, we need to update to 1.22.0
<cehteh>bleh ... tried to add with nm-appplet gives me 'insufficient privileges'
<bricewge>Same issue here, never managed to make it work without sudo :(
<cehteh>lol .. why did you recommend it then :D
<cehteh>my experiences with network manager are not really stellar ... okish for causual laptop use but if possible i try to configure things in the system
<cehteh>did you try nmcli? i may do that next
<cehteh>too bad i am leaving tomorrow for longer travel, would like to have wireguard working by then, taking only the guix laptop with me
<bricewge>You can't configure route without writing a custom service, that's why
<cehteh>lemme try running nm-applet as root
<cehteh>ok doesnt work, no vpn, other things to do
<dstolfa>networkmanager works really well under redhat systems, but it has always failed me under anything else
<sailorCat>Hi, I somehow got an error "Symbol’s function definition is void: make-closure" when I restart the emacs-next package after few attempts I decided to get a new fresh installation. But seems `guix package -r emacs-next && guix package -i emacs-next` didn't do it for me. What is the way to fully reinstall a package in guix?
<mfg>I can't help you, but the emacs-next package is almost one year old
<mfg>that is the commit it uses is from August 31 2020
<sailorCat>shame
<minikN>Hello, does someone have instructions on how to install composer (php) with guix? I installed it manually according to https://getcomposer.org/download/, I also installed the offical php package in a profile. But I can't execute composer: /usr/bin/env 'php': no such file or directory. Is there a better way to install php & composer?
<mfg>minikN: if you installed it in a profile, have you loaded that profile?
<minikN>Yup. `which php` returns correct path.
<mfg>Hm i don't know php that much, but i ran the instructions on the composer page and obtained composer.phar
<mfg>i can execute it and it gives noerrors
<mfg>did you try a specific command?
<mfg>i am inside `guix environment --ad-hoc php`
<minikN>Okay, never mind. Problem is not Guix, it's my IDE. Sry.
<mfg>No problem :-)
***Guest5707 is now known as roptat
<roptat>hi guix!
<GuixNoob>Hi!
<GuixNoob>Can we ask questions relating to package writing here ? :)
<MysteriousSilver>sure :D
<GuixNoob>Cool ^^
<GuixNoob>So I am trying to package the signal desktop app (wich is under AGPL and can be found here https://github.com/signalapp/Signal-Desktop) but I struggle a bit with choosing the adequate build system, how to you choose between all existing?
<rekado>GuixNoob: I’m afraid this will be terribly difficult to package.
<rekado>the JavaScript world is at odds with scalable approaches to packaging.
<MysteriousSilver> http://dustycloud.org/blog/javascript-packaging-dystopia/
***GuixNoob is now known as hlecomte
<hlecomte>How terribly difficult ?=L
<hlecomte>like nearly impossible ?
<rekado>at least “nearly impossible” to do well
<rekado>there are importers (that curiously still haven’t made their weird into Guix proper) for NPM packages
<rekado>but you’d end up with hundreds (or thousands) of packages.
<hlecomte>sad :/
<rekado>a simple example is this: https://github.com/guix-science/guix-science/blob/master/guix-science/packages/rstudio-node.scm
<rekado>these are the runtime npm dependencies for *one* package.
<hlecomte>X)
<abrenon>roptat: hi
<roptat>abrenon, hi! I forgot about you, I was busy packaging android stuff ^^'
<abrenon>you forgot about me ?!
*abrenon is offended : )
<roptat>at least, now I'm one dependency away from packaging adb :)
<abrenon>I just wanted to make sure I hadn't missed some feedback requiring to redraft the patch
<roptat>but that dependency requires three more java packages :(
<abrenon>that's great news !!
<abrenon>milliseconds, hey ?
<abrenon>well good luck with those
<roptat>so, I probably decided on the wrong approach. I basically reimplemented the whole build system in guile, without making the same assumptions that we're building all at once from the complete source tree
<roptat>I fixed many issues I had with the cc_* type of targets, but now I need to implement the whole build commands and option parsing for the java_* targets
<abrenon>reimplement the whole adb build system ?
<roptat>the whole android build system, actually
<roptat>(called soong)
<abrenon>that sounds like a lot
<abrenon>how come nothing can be reused ?
<roptat>maybe it can be, but it makes so many assumptions on what can be used from the system, on where build artifacts go, on what has been checked out...
<roptat>you can't use soong directly, it's not meant to be split in multiple packages
<abrenon>hmmm
<hlecomte>rekado Wow something like dependencies for just one package?! That's insane .. So there is no workaround to package signal on guix ? :/
<abrenon>roptat: and why did you called your approach "wrong" ?
<roptat>it's not "wrong", but it's definitely a lot of work ^^
<abrenon>it looks so yeah
<roptat>that's the part that takes care of cc_* targets, mostly: https://framagit.org/tyreunom/guix-android/-/blob/master/android/build/soong-cc.scm
<roptat>this is the whole build system: https://framagit.org/tyreunom/guix-android/-/tree/master/android/build
<roptat>instead of using soong, I parse the Android.bp files and generate a Makefile
<abrenon>^^
<rekado>hlecomte: this *is* the workaround
<abrenon>is that the make-makefile-rule and friends ?
<rekado>you’d import these hundreds of NPM packages and then move on.
<roptat>yeah, in a blueprint file you have different type of targets (like cc_library, cc_binary, ...) and each target has its own rules. So soong-cc implements them by generating the right makefile-rule objects, given the configuration and type, then the soong-build-system knows how to generate a Makefile from the and run make
<roptat>for instance, this is where I generate the rules for building sources to a .o object: https://framagit.org/tyreunom/guix-android/-/blob/master/android/build/soong-cc.scm#L576
<roptat>that's common to every cc_* targets
<roptat>but depending on the actual target, I'll use a different rule to use these .o objects: https://framagit.org/tyreunom/guix-android/-/blob/master/android/build/soong-cc.scm#L856
<roptat>and different install rules
<abrenon>I don't get l.576, it looks more to me like you're filtering .o files out of some list
<roptat>yeah, it's all the lines around, sorry: I'm looking at the extension of the file and apply build rules to different types
<roptat>like .proto, .c, .cc...
<roptat>oh, gitlab doesn't know about #;
<abrenon>that's quite impressive in any case
<leoprikler>so we will have a Scheme implementation of soong at some point?
<roptat>yeah, it's the dark magic that allows me to have nice and small packages: https://framagit.org/tyreunom/guix-android/-/blob/master/android/packages/android-tools.scm
<roptat>that's the crazy plan ^^'
<leoprikler>Wait until Google hears of this, steals the idea and makes it infinitely worse :P
<abrenon>isn't #; and other #.* specific to guix and not guile ?
<roptat>rekado, since you're here any comment on ant-bootstrap: https://issues.guix.gnu.org/49990 ?
<roptat>abrenon, ah it seems so, but it's supported by vim for instance, and it has no way to know it's guile and not something else
<roptat>mh... it's not even in the guile manual ^^'
<roptat>it's part of r6rs
<abrenon>ok I just confused #; with the #~, #$ and #$@ documented in the manual page about G-Expressions sorry
<roptat>g-expressions are specific to guix
<roptat>#; is to comment an s-expression
<abrenon>ahhh
***apteryx_ is now known as apteryx
<roptat>it's handy, you can comment without changing lines, like (+ 1 #;(compute-something) 3) -> 4
<abrenon>that's great ! so they're kind of sub-inline comments
<roptat>yeah, but they can also span multiple lines if you comment a complex s-expression
<abrenon>so, just comments-as-any-syntax-construct : ) cool
<roptat>so, I'm still missing code for handling java targets, I could create the packages for fastboot, I need to work on the importer so it's easier to recreate/reimport the packages each time there's a version change that could change the dependencies, and I should take the targets (host/device mostly) into account
<roptat>I'm still not entirely sure if the difference is simply the target triple, or if there can be different targets for the same triple
<abrenon>hmmm so "only one more package to go" but still a lot of work I see : )
<raghavgururajan>Hello Guix!
<Gooberpatrol66>Is there a way to force grub-efi-bootloader to install to a specific partition? It keeps installing to the wrong one.
<rekado>roptat: the behavior shouldn’t change just because you’re printing something.
<roptat>rekado, I know, but it does...
<roptat>unless the substitution does something else
<muradm>Hello Guix
<slyfox>is it just me or 'guix refresh' is broken on 'core-updates' branch? Tested as './pre-inst-env guix refresh libffi -u'. Works fine on 'master' branch.
<slyfox>The error is 'guix refresh: error: mkstemp: Permission denied'
<slyfox>Ah, I forgot to apply locally https://issues.guix.gnu.org/48434 to 'core-updates' ('master' has it for a long while)
<podiki[m]>anyone also run into issues with dbus and multiple profiles? https://issues.guix.gnu.org/48538
<podiki[m]>I just separated all my packages into profiles, so default is empty. I added glib to a profile to make sure I got XDG_DATA_DIRS exported from that profile to pick up dbus files, yet still seems you need dbus-run-session
<raghavgururajan>podiki[m]: You may have to run dbus-daemon as user, via dbus-run-session, so that dbus-enabled apps can communicate with the daemon.
<podiki[m]>raghavgururajan: does that entail modifying the default dbus service then?
<podiki[m]>or can I just do that on top of system dbus?
<raghavgururajan>dbus-service as in config.scm? If so that's system-level dbus-daemon. For some reason, apps run by user doesn't communicate with system-level daemon.
<raghavgururajan>just a sec.
<raghavgururajan>Are you using shepherd user-services?
<podiki[m]>dbus-service is in %desktop-services, so yes that would be the system-level one, but I guess you mean running a dbus user one too?
<raghavgururajan>Yes,
<podiki[m]>mostly desktop-services with some modifications (udev rules, syncthing, pcsd, etc.), but nothing dbus related I think
<raghavgururajan>you can add the following to .bash_profile
<raghavgururajan>if [[ ! -S ${XDG_RUNTIME_DIR-HOME/.cache}/shepherd/socket ]]; then
<raghavgururajan> dbus-run-session shepherd
<raghavgururajan>fi
<raghavgururajan>You can replace shepherd with anything else, but shepherd is a safe bet.
<podiki[m]>thanks, will try that
<raghavgururajan>What that snippet does is, when you login as user, the dbus-daemon session is started (along with shepherd) and exports correct variables. That dbus daemon runs as long as shepherd is running.
<raghavgururajan>*started at user-level
<podiki[m]>hrmm, doesn't seem to help. do I need to do anything for applications to talk to the user dbus daemon instead of system one?
<str1ngs>raghavgururajan: hello I wanted to test g-golf with gtk4 is there a staging branch for gtk4?
<raghavgururajan>Didn't work after restart?
<podiki[m]>yeah, just restarted. my test case is running a flatpak applications, as that needs to talk to dbus on startup. it complains it can't find the service file it wants unless I use dbus-run-session
<raghavgururajan>just a sec
<podiki[m]>thanks! (i'm also learning more about dbus daemons now...)
<raghavgururajan>str1ngs: There is no separate branch for gtk4, but wip-gnome branch is used for gtk4. There updates in that branch that is required by gtk4. The patches in #48554 are to applied and developed on that branch.
<raghavgururajan>podiki[m]: What's the output of `env | grep DBUS`
<podiki[m]>hrm, maybe I want to do a dbus-launch for my xsession (using the default gdm, but my own .xsession script to launch a WM)
<str1ngs>raghavgururajan if I'm reading this I can use wip-gnome branch and apply the patches from #48554?
<podiki[m]>GDM_DBUS_DAEMON=/gnu/store/vjsdrzl14djib9sbzc1lbd6iwg5ivgk5-gdm-dbus-wrapper
<podiki[m]>DBUS_FATAL_WARNINGS=0
<podiki[m]>DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-kBKnuzZDpG,guid=d22c7a86bd811807f289f661611a9ac7
<str1ngs>raghavgururajan for context I'm looking at contributing not using this as an end user :)
<raghavgururajan>str1ngs: Yep!
<str1ngs>raghavgururajan great TYVM kind sir.
*str1ngs dusts his off emacs debbugs-gnu
<raghavgururajan>podiki[m]: The variable "DBUS_SESSION_BUS_ADDRESS" is exported by that snippet, which is good.
<raghavgururajan>podiki[m]: For flatpak, you may have to extent the XDG_DATA_DIRS to flatpak's share-dir.
<cossidini>im trying to get stumpwm to run on aarch64 with guix but the build fails for sbcl. the ci server shows that sbcl build fails too. is there a way to check if it has ever successfuly been built?
<raghavgururajan>Just a sec.
<podiki[m]>raghavgururajan: yes, I made sure glib was in the profile to get XDG_DATA_DIRS. in comparison, when all was in one default user profile this worked. so it is a profile separation thing
<podiki[m]>I don't see anything that would appear in /etc/dbus-1 (except pulseaudio) from my desktop profile. I see the flatpak dbus stuff in the share folder of the profile and in my XDG_DATA_DIRS
<raghavgururajan>podiki[m]: Regarding multiple profiles, do you have this (https://paste.debian.net/plain/1207884) in your .bash_profile?
<podiki[m]>yes (and running zsh, for what that is worth, so it is in my .zprofile; but env does look to pick up everything from the profiles)
***niko is now known as o
<podiki[m]>is there something still hardcoded in dbus itself or the configuration? or is this just getting dbus-daemon/launch correct?
<cehteh>meh .. no actual rust in guix either :/
<rekado>of course there’s rust
<cehteh>talking about 1.53 at least
<cehteh>'actual' ... we are at 1.54
<raghavgururajan>podiki[m]: What's the output of `env | grep XDG_DATA_DIRS`
<cehteh>1.53 came with some significant changes
<cehteh>and rustup doesnt work either
<rekado>cehteh: have you considered packaging the new versions?
<cehteh>rekado: yes
<podiki[m]>`XDG_DATA_DIRS=/home/john/.config/guix/profiles/desktop/desktop/share:/run/current-system/profile/share:/home/john/.guix-profile/share:/run/current-system/profile/share`
<raghavgururajan>podiki[m]: Its missing /var/lib/flatpak/exports/share
<rekado>cehteh: is there a bug report about rustup?
<podiki[m]>where the .../profiles/desktop/desktop is the profile with all the desktop apps, and glib
<cehteh>rekado: its rather by design
<cehteh>rekado: that downloads a binary blob into /tmp and executes it, which wont be working in guix for now
<podiki[m]>raghavgururajan: yes, as well as the .local/share.. place flatpak wants to look, but that's for apps installed there to appear in eg launchers. I can add what it wants just in case (though /var/.. doesn't exist on guix of course)
<Noisytoot>Has anyone here ran Guix inside VMM?
<cehteh>rekado: i am good at complaining but bad at scheme/packaging and now i have no time to change that yet
<cehteh>rekado: fyi https://www.reddit.com/r/GUIX/comments/hjndh2/rustup_rls_in_guixsd/
<rekado>I don’t use reddit for bug reports
<rekado>would be better to have it in the bug tracker
<cehteh>thats why i saied 'fyi' not meant as bug report
<rekado>yes, I understand.
<rekado>would be better to have a bug report
<cehteh>but filing a generic bug "cant run arbitary executables downloaded from the internet" against guix would be pretty pointless i guess
<podiki[m]>raghavgururajan: no change. the dbus service file it wants is in XDG_DATA_DIRS (the share/dbus-1... as expected). something is not looking for dbus stuff besides in default profile
<podiki[m]>while dbus-run-session picks it up...
<cehteh>besides i dont like the curl foo.sh | sh installer idea its pretty broken i am ok with it not being fixed
<cehteh>was just some last resort try
<rekado>cehteh: I don’t understand the initial complaint then. You said “and rustup doesn’t work either”.
<rekado>¯\_(ツ)_/¯
<cehteh>would be nicer to have some way to upgrade packages in guix with a fingersnip ..
<raghavgururajan>podiki[m]: Ah, it might be issue you mentioned earlier. dbus is looking at only default profile. You might be right about that.
<rekado>oh well. I don’t really want to work on rust.
<podiki[m]> https://issues.guix.gnu.org/48538 I believe is it
<podiki[m]>seems like your idea of running a user dbus-daemon should get around it though
<raghavgururajan>What app you had the issue with?
<cehteh>i am currently dont even know where to start when i want to repackage it .. can i guix edit rust save the file $somewhere (is it really edit? or just 'view') and then at least for me build and install it before sending a patch?
<cehteh>well sory i am nagging about this, i coudl read the docs, but i am in a hurry packing stuff leaving tomorrow by train
<podiki[m]>raghavgururajan: seems to mainly affect Element (matrix client). It wants the flatpak portal to run on startup and wont' find it without dbus-launch
<raghavgururajan>podiki[m]: I have a confusion. Was it the app installed by flatpak or installed by guix in one of the extra-profiles, that doesn't coonect to the dbus?
<raghavgururajan>Ah.
<podiki[m]>flatpak is installed in the desktop profile (my default profile is now empty), and app is installed by flatpak and run by flatpak. flatpak has some dbus stuff it wants to launch (e.g. to open links or such)
***ec is now known as tl
<podiki[m]>but based on that issue I linked earlier, this seems to be a general problem if your dbus services are not in the default profile
<podiki[m]>so I guess I can investigate why system dbus doesn't use xdg_data_dirs (which should get dbus from all profiles), or somehow running a user daemon that is then used by programs instead of using the system one
<iskarian>good morning guix :)
<sneek>Welcome back iskarian, you have 2 messages!
<sneek>iskarian, yoctocell says: I saw your patch for the `git-fetch' updater, and it worked great when I tested it on a few packages hosted on GitHub.. Thank you for the work! :-)
<sneek>iskarian, yoctocell says: I think the next step would be to add a `generic-git' updater, so that it works on any Git repo, and not just ones on GitHubl
<raghavgururajan>podiki[m]: Could you try removing element by `flatpak uninstall im.riot.Riot --delete-data` and re-installing as *user* via `flatpak install --user flathub im.riot.Riot`?
*raghavgururajan tries to reproduce the problem.
<podiki[m]>it is installed via --user (don't think you can otherwise without sudo)
<podiki[m]>I should remove the dbus shepard lines and see what the default gdm dbus env is, to compare; but seems like launched by system is not respecting xdg_data_dirs (or is launched earlier and only with default like /etc/ to look at?)
<podiki[m]>my other guess it so try another WM (I just have awesome installed) to see if it is how I launch my WM messing with dbus. but again, worked without any workarounds when all was in same profile, so pretty sure that is the issue
<podiki[m]>(also going to investigate dbus and xdg_data_dirs more broadly)
<raghavgururajan>podiki[m]: When you remove dbus shepherd line, let me know if "DBUS_SESSION_BUS_ADDRESS" is still exported in env.
<podiki[m]>raghavgururajan: without shepherd line:
<podiki[m]>`GDM_DBUS_DAEMON=/gnu/store/vjsdrzl14djib9sbzc1lbd6iwg5ivgk5-gdm-dbus-wrapper
<podiki[m]>DBUS_FATAL_WARNINGS=0
<podiki[m]>DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-rBOGo0Ceu4,guid=6aa3e2850b82c6b16c6cd2ae611aa7aa`
<podiki[m]>I'm going to try manually putting a dbus in how I launch my wm, let's see if that does anything
<raghavgururajan>Cool!
<podiki[m]>raghavgururajan: with manual dbus-launch --exit-with-session my-xmonad-bin
<podiki[m]>`GDM_DBUS_DAEMON=/gnu/store/vjsdrzl14djib9sbzc1lbd6iwg5ivgk5-gdm-dbus-wrapper
<podiki[m]>DBUS_FATAL_WARNINGS=0
<podiki[m]>DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-zQRocAFaEU,guid=8915d523e3c7fa276915e79f611aa834`
<podiki[m]>and that works
<raghavgururajan>Great! :)
<podiki[m]>I'm guessing that is a work around similar to using dbus-run-session
<raghavgururajan>Yeah, dbus-launch also starts a dbus-daemon at user-level.
<podiki[m]>(I still have another issue with it not picking up the xdg-desktop-portal service files correctly, but that was true before)
<raghavgururajan>xdg-desktop-portal service?
<podiki[m]>yeah, as in here https://issues.guix.gnu.org/49972
<podiki[m]>seems to be something gnome probably does that I don't with just running a WM
<raghavgururajan>Nice!
<podiki[m]>looks like I'll have to dig a little in what gnome does for dbus still
<podiki[m]>thanks for the help on dbus raghavgururajan! maybe it will help me figure out my xdg-desktop-portal issue now that I know a little more
<podiki[m]>I wonder if dbus-update-activation-environment would also have been useful to try
<raghavgururajan>podiki[m]: No worries!
***john__ is now known as gaqwas
<raghavgururajan>podiki[m]: Question, were you starting X without using display-manager?
<podiki[m]>I'm using gdm (as set by default by guix installer), but made a ~/.xsession that symlinks to my .xinitrc (executable) that normally ends with something like `exec /path/to/my/xmonad`
<podiki[m]>that last line I changed to be `dbus-launch --exit-with-session /path/to/my/xmonad`
<podiki[m]>I think this is not the right way to add a custom session to gdm, so I should fix that
<podiki[m]>I also just noticed at-spi-bus-launcher starts at startup by default (the only thing I think, must be from gdm), in case that means anything
<raghavgururajan>podiki[m]: I have trying to start X manually without display manager. Failed multiple times.
<podiki[m]>I haven't tried on guix (just used gdm), but in the past just used startx with an .xinitrc (last line being the 'exec wm')
<Noisytoot>Why does GRUB not output to the serial console with this config.scm (when running in QEMU)?: https://bpa.st/JCNQ
<podiki[m]> https://github.com/podiki/dot.me/tree/master/x11 my dots, though I need to upload the guix branch (will do later)
<iskarian>yoctocell, thanks for testing, and for sharing your generic-git updater! I'm not sure how I feel about using properties, or at least requiring properties when we can guess - I'd rather it enable support without editing packages where possible
<iskarian>And if I'm reading it right (I'm not super familiar with (guix git)) this updater has to clone the repo for every package it's checking?
<raghavgururajan>podiki[m], Thanks!
<ruffni>do i understand the docs correctly that ALSA+PulseAudio is the default? but there's no shepherd service for those, right?
<justin>any 'guix deploy' guru's that can help with deploying from x86 to aarch64?
<sneek>Welcome back justin, you have 1 message!
<sneek>justin, marusich says: It's possible to build an ISO-9660 image for GuixSD, but since the feature was added recently, a pre-built image isn't available yet. To build your own, run 'guix system disk-image -t iso9660 gnu/system/install.scm' from the root directory of the Guix source tree.
<podiki[m]>@ragh
<podiki[m]>raghavgururajan: updated with mostly current guix config related stuff https://github.com/podiki/dot.me/tree/move-to-guix
<podiki[m]>raghavgururajan: I'm still stuck on the xdg-desktop-portal dbus refusing to find some of it's service files, but no leads. I'll probably make my xsession cleaner and look at what stuff gnome does with dbus. very odd though
<muradm>hi guix, any one could check http://issues.guix.gnu.org/49969 ?
<acrow>I am running GuixSD, pulled this morning from master. I've installed the gramps package and whenever it is started I get the following error message, "grampsgui.py: line 538: GTK translations missing, GUI will be broken, especially for RTL languages!". This is just informational but it bugs me. I have no other examples of this happening but I keep thinking I'm not installing something else I need like, and I've already done this one,
<acrow>shared-mime-info. Running en_US and I've specified (operating-system ..(locale "en_US.utf8")...). Anyone else have other ideas?
<acrow>I see that the ubuntu community solves this same problem by installing the package 'language-pack-gnome-en'. Do we have an analog to that that I'm missing?
<acrow>Of course I'm running the xfce window-manager and not gnome.
<acrow>I think you are correct ruffni that PulseAudio is a default and nothing else is needed from shepherd but you might want to be sure you put yourself in the 'audio' group to have it work as other than root.