IRC channel logs


back to list of logs

<methalo_>I'm trying to replace 'HICOLOR_ICON_DIR=`$PKG_CONFIG --variable=prefix gtk+-3.0`/share/icons/hicolor' in the configure file
<methalo_>seems to have no effect
<enderby>hi all, have a guile question, running guix on a foreign distro. when i do 'guild compile <file>', it writes to $HOME/.cache/guile... and not the directory given with (%site-ccache-dir) in repl. is there a configuration file/setting i have to tell guile where to compile to?
<k`>Do gtk+ and mate-themes have to be in sync? I want to update gtk+, but mate-themes lags behind gtk+ releases by a bit.
<efraim>k`: it's OK, we don't have all of mate packaged yet and they've moved to full gtk3 support
<k`>efraim: Alright. I'll push the gtk+ update to staging and the mate-themes update to master.
<k`>Aw snap. I didn't press "t" and "o" hard enough and now the gtk+ update log says "Update 3.22.12." :(
<k`>The Thinkpad keyboard's bounciness spoils me.
<k`>marusich: Hi!
***Guest26031 is now known as sturm
<lyss>alright, time to hack around with guixsd in a vm
<lfam>Welcome! Feel free to ask here or on <> if you need help with anything
<lyss>lfam: actually I've been idling here for a week :p
<lfam>Cool :)
<lyss>lfam: actually, I do have a question.
<lyss>Why does invoking guix install consistently crash X under Debian? o_O
<lfam>lyss: Hm, I'm using Debian, too. What do you mean by `guix install`? That's not a Guix command
<lyss>correction: logs out the entire system
<lyss>(I tried it without X)
<lyss>anything that involves installing packages
<lfam>Hm, weird! I've never experienced that before
<lyss>no kidding :p
<lfam>What version of Guix are you using?
<lfam>And what version of Debian?
<lyss>0.12.0 apparently
<lyss>and stretch
<lfam>Do you see any output from the command before you get logged out?
<lyss>yeah, it was in the middle of isntalling
<lyss>(I tried 2 different invokations, same thing at differen points)
<lyss>and it did what I assume was correct (downloading, installing, rinse and repeat)
<lyss>and then all of a sudden... poof
<lyss>no errors I can see
<lfam>And what about `journalctl --unit guix-daemon`? Anything interesting from `journalctl -a`?
<lyss>"unexpected Nix daemon error: interrupted by the user"
<lyss>and "spurious SIGPOLL"
<lfam>Those are typical: the build was interrupted
<lyss>so.. the bug is that there is a bug? :P
<lyss>seems awfully circular haha
<lfam>I mean, those messages show up under normal usage. So I think you should inspect the full system log with `journacltl -a`
<lfam>I mean, `journalctl -a`
<lyss>i did, nothing relevant there
<lfam>Nothing at all? Just normal thing and then your user session ends?
<lyss>all the suspicious stuff is from the session ending :p
<lfam>Very weird. I would also check the kernel logs and, as a last resort, strace the guix-daemon
<lfam>If you log strace to a file, you should be able to see the last thing it does before you get logged out
<lfam>Maybe it's consistent
<lyss>lfam: nothing out of the ordinary from dmesg
<lfam>I'm using Sid, not Stretch, so I can't really attempt to reproduce.
<lyss>lfam: Why for the love of God would anybody ever use Sid? :p
<efraim>my debian boxes normally run sid, with apt-listbugs its never really been a problem for me
<lfam>Stretch gets bug fixes last, so I don't like to use it
<lyss>hmph :p
<lyss>ACTION is surprised she isn't being flamed about not-using-Trisquel
<lyss>lfam: So, strace went down along with the rest of the system obviously
<lyss>but I can post the log anyway if you think it'd help
<lfam>I'd try running it as root in another TTY, unless the log looks complete
<lyss>it's stopped midline heh
<lfam>Since the daemon runs as root, it shouldn't be affected by your user getting logged out
<lyss>wait, how do I strace something from another user?
<lfam>You can pass the process PID to -p. I do it like this: `strace -f -p $(pgrep guix-daemon | head -n1)`
<lyss>oh i see
<lyss>lfam: That log seems to die *because* the originating guix process died
<lyss>Which doesn't really help :p
<lfam>Right, but I'm interested in where it stops :)
<lyss>Alright, I'll send the whole thing then :p
<lyss>enjoy :p
<lfam>Hm, stumped so far
<lyss>...quit so soon? =P
<lfam>lyss: Line 11970 stands out: [pid 6713] kill(-1, SIGKILL) = 0
<lfam>Maybe there is something interesting before that
<lfam>I'm not sure what context that gets executed in
<lfam>lyss: Are you using an existing installation of Stretch, or is it a VM image I can try out?
<lfam>efraim: Can we update 'less' on the master branch?
<lfam>And also libsigsegv?
<lfam>And libiconv too?
<Apteryx>So, no one else experience a failure to build the nss package? (except Marius who opened a bug regarding nss failure but for armhf (I'm on x86_64))
<lfam>Apteryx: It worked for me, but Marius reverted the update
<Apteryx>OK. Thanks for the info :)
<lfam>lyss: I'm stumped. Can you send a message to <> with as much information as you think is relevant? Like, the architecture you're using, OS name and version, Guix version, a command that triggers the bug, and the strace log?
<lfam>Apteryx: We are constantly having problems building that packages. It's really frustrating!
<lyss>tell lfam okay, will do it tomorrow if I remember
<Apteryx>Do we have GNU Ring in the repo? When I do guix package --search="GNU Ring" it returns pages of results.
<lyss>(is there formal syntax I messed up? oh well)
<Apteryx>Looks like we don't have it! I'll put this on my list ;)
<sumaru>trying to install guixsd and getting an error
<sumaru>wicd-1.7.4.drv fails to build
<sumaru>EOFError: EOF read where object expected
<sumaru>that looks to be in giscanner/ line 39
<sumaru>or maybe g-ir-scanner line 65
<sumaru>I don't even need wicd on this machine if there's an easy way to keep it from being included.
<sumaru>is it pulled in from %desktop-services or gnome-desktop-service anyone know?
<sumaru>k, it worked without gnome-desktop-service
<civodul>Hello Guix!
<sumaru>ok, now it worked with everything
<sumaru>either something changed in the last half hour, or you have to do this in a particular order
<Petter>Trying to update, but I get this error: "guix substitute: error: corrupt input while restoring '/gnu/store/srid31mxf6bz7js9ql7szn0x7izmklp5-mysql-5.7.18/bin/my_print_defaults' from #{read pipe}#"
<Petter>Not sure how to proceed. I've gotten this multiple times over some hours.
<brendyn>Petter: try --fallback
<sumaru>Petter: I get that constantly and I don't understand why.
<sumaru>it says something about network issues which I'm pretty sure (at least on my end) is not the problem
<sumaru>--fallback usually does work though
<brendyn>maybe guix gc to clean everything out?
<Petter>Yes, it suggests it could be a network issue as well, but I don't think so in this case. And it always fails on the same item.
<Petter>I'm suspecting a shared problem, so maybe we should fix it for everyone if possible.
<Petter>sumaru: Do you get this for mysql, or other packages as well?
<sumaru>I haven't noticed any particular package - Happens almost any time I upgrade or install something.
<civodul>Petter: unfortunately it's likely a problem on our server, but --fallback will indeed work around it
<Petter>Ok, I'll do the work around then.
<sumaru>how do I install a specific output of something again?
<sumaru>was it like transmission:gui or something, I haven't done this in a while
<sumaru>looks like yes
***boegel|quassel is now known as boegel
***roo is now known as Guest49156
<catonano>How is Hydra ? It didn't reply to a request of mine
<catonano>Ah I see only now that Petter had problems too
<Petter>Yeah, mysql appears broken.
<catonano>I did "guix environment guix" and it started building Perl. It's my first time
<catonano>I tried to interrogate it with the Emacs tools and it didn't reply
<catonano>Probably some recent updates triggered thhe rebuilding of fuundamental packages
<catonano>so now it's grinding
<Petter>Oh Panic. My system crashed while upgrading, it's most likely corrupt now :O It's a miracle it even booted.
<Petter>Wait, I'm running GuixSD!
<Petter>ACTION puts on sunglasses
<Petter>All is cool!
<civodul>Petter: :-)
<methalo_>Manjaro uses elogind,
<civodul>methalo_: nice!
<civodul>so they registered the "elogind" name on GitHub, right?
<civodul>should we switch to that, what's the story?
<methalo_>on GitHub page says: forked from wingo/elogind
<methalo_>civodul: I think it's a fork now, in the README the link to contribute is
<civodul>methalo_: ok, i hope it will evolve in a direction that is good both for Manjaro and GuixSD
<paroneayea>whoa holy shit exwm
<paroneayea>emacs based wm in guix what
<Sleep_Walker>I used it for a month
<Sleep_Walker>it was quite nice
<Sleep_Walker>with proper configuration like EDITOR=emacsclient it felt like there is finally fast terminal emulator in emacs
<Sleep_Walker>I had only one key binding for window split
<Sleep_Walker>(instead of i3/dwm one, tmux one and emacs)
<civodul>i wonder what happens when you check mail in Gnus :-)
<civodul>that makes me skeptical about exwm
<apteryx[m]>Hello Guix (from Matrix /w Riot client)
<Petter>Hello apteryx[m] :)
<ng0>do you think we could roll out the next release with an .iso for GuixSD? I don't even know what the status of the iso is other than someone looked into it
<ng0>was there some chat about iso with uefi+bios support?
<apteryx[m]>I'm on a quest to find a good IM software that does video/audio and support strong encryption. So far I've investigated GNU Ring and Riot. Next I'd like to try Tox and PSYC.
<ng0>good look finding a video/audio client for PSYC. PSYC is just the protocol.
<brendyn>I have used video chat with Qtox
<quiliro> can i make a mirror for all guix packages?
<civodul>quiliro: the easiest way is to run 'guix publish' on a machine that has the packages of interest in store
<apteryx[m]>brendyn: Was it a good experience?
<brendyn>Yes it worked just fine. I don't think the audio has noise cancellation though
<brendyn>feedback, i mean
<apteryx[m]>ng0: I see!
<apteryx[m]>brendyn: OK! From what I've read about Tox, I like that it appears to be streamlined and is pure p2p.
<lyss>brendyn: apteryx[m]: I would've used it, but it was really finnicky and didn't have a security audit, etc.
<brendyn>I don't know how much work is being done on it atm, there seems to be this strangely immature userbase around it, but Qtox is already quite functional it seems
<lyss>IIRC, my issues with them were:
<lyss>* IDs changed too much
<lyss>* IDs were confusing in general
<lyss>* Polished interface but rough around in the edges (a lot of subtle 'workaroundable' bugs IIRC)
<lyss>* Clients not bundled in Debian iirc
<lyss>* No security audit, despite making some big claims
<lyss>* Some drama around something or other (I don't remember what this was about)
<lyss>Any one of those I could tolerate, but... together it really didn't justify switching workflows.
<brendyn>I was playing with GNUnet before and after 20 minutes it connects to 50-100 nodes and maxes out 1 core of my cpu. i tried hosting a file and downloading it from myself as a test and found it was really slow
<lyss>hooray :p
<brendyn>"The Tox community is absolutely toxic, no pun intended. I have abandoned that ship.
<brendyn>" -- davexunit
<apteryx[m]>Yeah, I've seen that on hackernews ;)
<brendyn>its nice when people use the same usernames everywhere
<brendyn>makes the internet less wild west
<apteryx[m]>I'm impressed by Matrix community so far, and their Riot client is really polished.
<lyss>apteryx[m]: I keep telling myself I'll start using matrix soon (TM) so i dunno :p
<sturm>I'm looking to disable a single test in a Python package. Am I better to add a patch that adds the "unittest.skip" decorator to the offending test, or override the "check" phase with a custom test command?
<sturm>Overriding check feels more immediately transparent, but also adds more visual noise to the package definition.
<bavier>sturm: the choice sometimes depends on the reason for skipping the test
<bavier>sturm: a custom phase override is preferred if the reason for the test skip if related mostly to the guix build process
<bavier>sturm: a patch or snippet is better in the case where the test itself is broken in some way that would be better fixed at the source level
<sturm>bavier: thanks, that makes sense
<bavier>sturm: something that you'd want any user of 'guix build -S' to see
<brendyn>hmm, riot requires phantomjs
<apteryx[m]>I guess that's bad? ;)
<brendyn>Well the maintaner of phantomjs stepped down
<brendyn>hmm, the build script for Parabola is complex and uses npm to download heaps of stuff during the build phase
<bavier>a chat client seems like a strange use of phantomjs
<brendyn>... and the build failed. oh well
<brendyn>Seems like it may be a "webapp", so even the desktop version involves running a whole browser
<apteryx[m]>Oh! Well there is always a plugin to weechat to get started with Matrix :)
<apteryx[m]>I've also read there is support for Emacs
<lyss>apteryx[m]: because weechat and emacs are the same class of software? :P
<lyss>(I joke. I'm seriously considering trying out emacs w/ evil-mode or whatever they call it now.)
<quiliro>civodul: yes, i understand that....but i would like to make a mirror of the whole is for offline installation
<Apteryx>lyss: Emacs is the gift that keeps giving ;)
<lyss>Apteryx: I know, but.... emacs :S
<quiliro>civodul: if it is not possible to make a mirror of the substitutes, it would be useful to make a repo of the sources...even of both
<lyss>Apteryx: I guess my issue is that my entire system is based around the UNIX philosophy. vim, mutt, irssi, tmux, and so on
<lyss>If I switched just one to emacs, I'd be compelled to switch all.
<lyss>Which.... inertia :P
<quiliro>lyss: switch just one what?
<lyss>quiliro: one of "editor, irc, email, browser, etc"
<k`>lyss: That's what happened to me. :-P
<mbakke>lyss: I have emacs with evil mode, but only use it for mail and text editing (gasp!)
<mbakke>and occasionally the calculator :)
<mbakke>it's only been two years though..
<civodul>quiliro: then you should use an nginx proxy to
<civodul>quiliro: the nginx config would look like this:
<quiliro>Guix spegulo servisto.-
<quiliro>1. subenŝarĝu GuixSD USB
<quiliro>2. guix package -i nginx
<quiliro>3. utilizu
<quiliro>ke plus mi devas fari?
<thomassgn>(not sure if this should be in the #guile channel) How would I get to see the source of guiles built-in i.e. (ice-9 popen)?
<daviid>thomassgn: better ask guile related quiz in #guile, but sources are installed in $prefix/share/guile/@version/
<bavier>thomassgn: or you can get a tarball easily with 'guix build -S guile'
<daviid>thomassgn: actually you may run this: guile -c "(display (%site-dir)) (newline)"
<avoine>ACTION note that one
<efraim>I have to get in contact with csan soonish, I got my hands on a macbook g4, one of those with a PPC
<efraim>I also got another Mac old enough to try libreboot and I think I have enough machines to make a go at x32
<efraim>ACTION is on a plane atm
<catonano>efraim: a naive question: what's the point of having libreboot working on old Apple Macintoshes ? Also, what does x32 mean ?
<daviid>avoine: for info you also have guile -c "(display (%global-site-dir)) (newline)" and guile -c "(display (%site-ccache-dir)) (newline)". also, one can use pkg-config to get these values as well
<avoine>daviid: guile will search for modules in all of those?
<avoine>except site-ccache that would be only for go file
<daviid>avoine: guile will search here: %load-path
<daviid>guile -c "(display %load-path) (newline)"
<daviid>avoine: also see add-to-load-path in the manual...
<avoine>I'll check that up
<catonano>where is objdump in guix ?
<snape>catonano: in gcc-toolchain
<slyfox>civodul: WDYT of and a followup email?
<civodul>slyfox: oh!
<civodul>slyfox: i'll look into it
<civodul>haven't checked guix-devel email yet
<slyfox>no rush, just making sure someone will read it at some point :) thanks!
<catonano>snape: thanks !