IRC channel logs
2024-01-01.log
back to list of logs
<lilyp>Sadly no. The important thing atm would be webkitgtk (all three variants) so that other stuff can be built easily. The gnome metapackage is obviously also something to look for. <lilyp>Should we default to submitting custom manifests for branches going forward? <efraim>got a request a few days ago to know more about my attempt at supporting x32 on Guix. Now I'm down a rabbit hole <efraim>I thought about making a manifest for rust-team. I have one with all the packages using the cargo-build-system, wasn't sure about making one with everything that depends on them <efraim>rust-sysroot for i586-pc-gnu works with rust-1.75 <lilyp>You would implicitly capture all of gnome through librsvg as well; I think it might be more worthwile to just do Rust stuff. (Although there are some fun things like Rust → Webkit → rust-webkit) <efraim>"just" everything using cargo-build-system should probably be fine for most things, but I'd still need to change the build target at some point to actually build out everything else <efraim>rust-sysroot for x86_64-linux-gnux32 works with rust-1.73 <meorly>do I understand it correctly that guix home is not intended to produce _multiple_ home environments? <meorly>lilyp: I read your discussion about multiple profiles on the ML but I am not sure I understood the implications <lilyp>indeed, guix home currently just creates one profile <apteryx>hm, adb stopped working to transfer photos from an aged phone. what do people use? It seems it should be possible via MTP, but nautilus doesn't auto-detect the device <apteryx>debug mode is still enabled on the phone <apteryx>oh, it just detected it now; perhaps I had to select 'file transfer' as operating mode on device <apteryx>still, reaching to /adb:: in emacs shows nothing <apteryx>perhaps adb in TRAMP stopped working for our emacs <lechner>Hi, is perl-dbix-class building on the substitute servers? <lechner>apteryx / my phones are ancient but nautilus always finds them. i have libmtp and jmtpfs installed via Guix Home, as well as gmtp <lechner>whoops you already figured it out. yes, you have to consent on the device <apteryx>it still doesn`t work from TRAMP/emacs though <A1ice>any loongson 2f user? any information about use guix on this device? <Franciman>oh ok, local-file accepts a #:recursive? boolean value that when set to true for flat files retains the permissions <apteryx>lechner: does adb work for you? from stock emacs, you visit /adb:: and it should show you a dired view of its root file system <rekado>lilyp: I’ve built webkitgtk and the gnome package from the gnome-team branch <rekado>building @2.40, -next, and -with-libsoup2 now <lilyp>we still have 2.40 somewhere? <podiki>so what's happening with berlin? build rate has been really low, mesa-updates going very slowly (just waiting for builds at this point) <apteryx>everything appears to be running smoothly, so I'm not sure <apteryx>perhaps something is hung up in cuirass? <apteryx>seems to be running without a glitch according to /var/log/cuirass.log <podiki>very few workers online right now it looks like <podiki>and global stats show big slow down in number of builds in last week or 2? (hard to say with lots of builds submitted but seemed to have stalled out a bit maybe now is back on track?) <rekado>the db queries to get a pending build are very slow <rekado>I’ve investigated this as time permitted but I haven’t found anything obvious <rekado>there’s definitely something wrong here, but I feel a little alone in figuring this out and I’m not expecting to make any progress. <cmiller>I am uptading Docker to 20.10.27 from 20.10.25. If I create a commit, does it only require docker and docker-libnetwork or docker-cli as well? <podiki>so building would be fine it is just slow at getting the next build to do? <cmiller>I am asking, since %docker-version is used by docker-cli as well but if I look in past commit logs for updating Docker, they did not include the docker-cli <rekado>podiki: way too slow to get the next build <rekado>the workers keep asking for work but don’t get anything <rekado>I don’t know if the delay in getting a response makes a difference here <apteryx>postgres doesn't seem to be constrained by CPU or disk <lilyp>very slow as in it took until rekado manually rebuilt webkit on gnome-team to get there (that's more than a week of stall) <avv>Hey. Question for Guix SD: if I want to, say, edit a file under /run/current-system/profile/xsessions, can I edit as usual or should it be done in the configuration script? If so, is there a resource/example on how it should be done? Thanks <podiki>avv: you can't edit that directly (it is in /gnu/store and is read-only) so that would be through your system configuration <avv>is it through 'extra-special-file' that I should be editing system files? I must have missed it while looking through the docs. Much appreciated, podiki and paenuts :) <podiki>generally no, it would be through system configuration; for xsessions adding the relevant packages (whatever wm/de) to your system configuration should be enough for them to show up if that's what you are trying to do <podiki>extra-special-file is rather manual and you can use that, but hopefully most of what you want to do is through exposed configuration already <avv>actually I'm just trying to run a script before starting a session in gdm. Gdm has x-session option, but I'm kind of lost when using it haha. I thought modifying the .desktop file would be easier, but now I noticed I have to basically find out in which directory my wm executable (dwm) lives in and modify the .desktop file accordingly haha. I think I'll try modifying gdm's x-session option then <avv>And oddly enough, when x-session is left as a blank function, gdm won't start. Do I need to handle gdm's initialization in there as well? <rekado>apteryx: very few builds are pushed out to build nodes. Many branches have been queued up for days without any progress whatever. <rekado>I just built the r-updates branch manually on ci.guix.gnu.org <rekado>there is no obvious problem with it, except that it looks like it took time off for the holidays <rekado>I find the SQL query time suspicious, but I don’t know if that’s indicative of an actual problem. <podiki>avv: might want to look at what things gdm will run, from a user's home directory (like .xinitrc, other dot files), I don't use gdm so i'm not sure but that might be more what you want <lalebarde>I have a package (firefox or firefox-esr) for which guix tries to compile `gkrust` (lib) and fails with error: `could not compile `gkrust` (lib) Caused by: make: INTERNAL: Exiting with 1 jobserver tokens available; should be 2! error: in phase 'build': uncaught exception: %exception #<&invoke-error program: "./mach" arguments: ("build") <lalebarde>exit-status: 2 term-signal: #f stop-signal: #f> <podiki>please ask in the channel for that package <lalebarde>Sorry, I am not in the right Libera channel. <podiki>but browsers tend to be very resource intensive and need a lot of ram/swap to compile (>16 gigs for icecat/firefox I believe) <apteryx>rekado: about how long were the queries you tested? <podiki>cuirass also only shows about 10 machines right now, and about half build slots idle (though tons of x86_64 jobs waiting) <podiki>(not sure if the log is for that current build or previous one, but either way is short and shows complete) <loopback___>is it possible to tell if a store item is a package, a tarball, etc.? Want to create an archive of source code for everything on an installation and the tools to build it <loopback___>I'm thinking I could try checking if the package name and version exists in a channel, then download the sources for it, but idk how to do it and there's probably other ways to do things <lilyp>do we have the newest python-tzdata somewhere already? <rekado>apteryx: (db-get-pending-build "x86_64-linux") takes about 10 seconds, and it returns rust things from the mesa-updates branch. <rekado>apteryx: for aarch64 it’s consistently 20s <rekado>and we’ve got no aarch64 builds happening <rekado>at first I thought there’s a problem with the connection between the head node and the builders, because we had maintenance scheduled in the data centre <rekado>there was in fact a faulty ethernet cable, but that’s gone now, and all three rack switches have been swapped for new ones, too <rekado>there used to be network related errors in the cuirass-remote-server log, but no more. <podiki>and the workers status showing few build machines is expected right now ?(maintenance? gc?) <rekado>we now have reliable connections, but the nodes are not being given anything to work on, in spite of a huge backlog <rekado>the nodes are running cuirass-remote-worker, which talks to the head node’s cuirass-remote-server <podiki>should i cancel the several big mesa-updates evaluations and try to restart them some hours later? <rekado>I don’t think that will do us any good <rekado>the cuirass-remote-server clears the number of workers based on some timeout <rekado>looking at node 101 now I see that it keeps pinging the head node <rekado>once for each of the four workers it provides <rekado>I just ran ‘(parameterize ((%package-database "dbname=cuirass")) (with-database (db-clear-workers)))’ to clear the workers table <rekado>just restarted cuirass-remote-server <rekado>lots of workers are back now but they don’t get any jobs <lilyp>We should probably rethink our databasing approach. Can we do semweb with guile-rdf? :) <lilyp>Welp, I caught up some of the gnome-team backlog at last. Someone still needs to close them issues though.