IRC channel logs


back to list of logs

<str1ngs>fusion809: yes, guix pull fails with ice-9/boot-9.scm:760:25: no code for module (guix profiling)
<fusion809>Aha yep that's the exact error I get. So use 6e11... commit instead eh?
<str1ngs>yes use that commit for now, let me know if that works for you aswell
<fusion809>Yep it worked
***pksadiq_ is now known as pksadiq
<vagrantc>anyone use an OpenPGP smartcard with guixsd? not sure what i need to get it recognized
<vagrantc> it's listed in usb devices, dmesg output recognizes it... but "gpg --card-status" can't find it
<atheia>Good morning guix
<atheia>has anyone else reported issues with Guix pull yet?
<atheia>to do with the profiling module?
<clacke[m]>thank you!
<clacke[m]>now I don't have to :-)
<clacke[m]>Tried with 0.13.0-14 and 0.14.0 for both daemon and client, and it barfs
<g_bor>Hello guix!
<g_bor>I see hydra failing with gateway timeout.
<g_bor>Do you also have a problem?
<nee``>g_bor: Are you trying to search via the webinterface? I also get that quite often. The timeouts are probably set too low. It usually works after a few retries.
<g_bor>No, just downloading soucre tarballs for builds.
<g_bor>I usually succeed using --fallback.
<g_bor>I'd like to now if we have a breamer theme or template to create presentations about guix and guixsd.
<g_bor>The Hungarian FSF team asked me to make and introductory session on guixsd, and I would like to create a beamer presentation.
<rekado>g_bor: same problem with
<rekado>Hi Guix!
<g_bor>It is really annoying..
<g_bor>Do you know what happened?
<rekado>I don’t understand the error report fully.
<rekado>so you run “guix build -S foo” and … what happens?
<g_bor>So, what happens it basically this:
<g_bor>Sorry, it seems that mirror works...
<g_bor>guix substitute: error: download from '' failed: 504, "Gateway Time-out"
<g_bor>I have this
<g_bor>when I use --fallback, then I can go on, but it takes a while to time out...
<g_bor>Now I see it for almost every download...
<g_bor>However it seems to work sometimes...
<g_bor>upstream downloads seem to work.
<g_bor>The first that was consistently failing for me was java-testng, but soon I realized, that it's not working for a lot of other packages...
<adfeno>Hi all! ;)
<g_bor>rekado: Could you please have a look at the way I changed the defaults around in my lat two commit on branch
<adfeno>I'm using Guix in a foreign distro. I tried to install mlt package, installation completed, but it seems I'm unable to use something like `melt video.webm'. I already have ffmpeg installed from GNU Guix, but doing `melt -query formats' tells me that melt failed to load the avformat consumer.
<g_bor>I think I've done nothing worng, but maybe changing the definition of ant this way is not wise.
<g_bor>rekado: it seems that the mirror is failing for me after all... I don't know why... Other things seem to work fine...
<rekado>g_bor: could you send these commits as patches to the list?
<rekado>g_bor: does work for you? I don’t know if something’s wrong with the mirror.
<g_bor>I'll check berlin.
<g_bor>Yes I will send the patches to the list.
<rekado>g_bor: thanks!
<rekado>civodul: looks like the profiling commits broke master.
<civodul>damn it
<rekado>g_bor: I also get 504 with :(
<civodul>hmm the bug report does not say how to reproduce
<rekado>when I do ’guix environment guix’ it aborts with:
<rekado>guix environment: error: build failed: derivation `/gnu/store/jgh9g2r5drasy320wg95cipjpdll9aih-guile-2.0.14.drv' has incorrect output `/gnu/store/06y7gah4s74w66v0ba8pfd0azhkdkxhw-guile-2.0.14', should be `/gnu/store/p50qidshrjvvd7d1955navjkijr2fg32-guile-2.0.14'
<g_bor>How can we find out what the problem with is?
<pmikkelsen>hello guix, i have a problem with my installation of guix. I am on guixSD and use a git checkout, but after todays git pull && make, i can't make a new environment. It says
<pmikkelsen>guix environment: error: build failed: derivation `/gnu/store/q9f9vy3a5dl9ysn6qp134wv7ih0saany-guile-2.0.14.drv' has incorrect output `/gnu/store/06y7gah4s74w66v0ba8pfd0azhkdkxhw-guile-2.0.14', should be `/gnu/store/p50qidshrjvvd7d1955navjkijr2fg32-guile-2.0.14'
<civodul>rekado: oh ok
<civodul>well let's just revert, i was mistaken on that one
<kristofer>yep the last commit I could get guix pull to build was 3bbc6fae15fe013aa39b88f0e1866a098fb936fd
<rekado>hmm, cuirass on the new berlin server still crashes because of missing drv files
<rekado>I already removed the cuirass.db and the cache.
<rekado>civodul: on berlin I’m running “guix pull” and it crashes with this error:
<rekado>ice-9/boot-9.scm:760:25: In procedure dispatch-exception:
<rekado>ice-9/boot-9.scm:760:25: no code for module (guix profiling)
<g_bor>rekado: I've just checked berlin, I can downlonad from there just fine.
<rekado>hmm, looks like the database is borked on berlin.
<rekado>I’m running “guix gc --verify=repair,contents” to fix it, but I don’t get far
<rekado>guix gc: error: build failed: executing SQLite statement: FOREIGN KEY constraint failed
<rekado>I’m going to start anew. “guix gc” as far as it goes, then rebuild everything from 0.14.0.
<civodul>rekado: bah
<rekado>I don’t understand how this could have happened
<civodul>rekado: the 'guix pull' issue should be fixed now
<civodul>what makes you think the db is corrupt?
<rekado>all I did was running “guix copy” on all first-level store directories to copy items from the old server to the new
<rekado>something odd happened to the db, because I couldn’t do “guix build -S gias-xcas”. It would complain that the derivation is missing.
<rekado>and it was missing.
<civodul>was the file missing, or was the db entry missing?
<str1ngs>civodul: looks like guix pull is fixed of the profile issue
<rekado>the file was missing, but why was there a db entry then?
<rekado>so I ran “guix gc --verify=repair”, but it failed.
<civodul>it should have removed the db entry
<rekado>it removed some db entries
<rekado>but then it tried to remove entries for built items
<rekado>and failed to remove them because they still had referrers
<rekado>and then this happened and caused guix gc to abort: guix gc: error: build failed: executing SQLite statement: FOREIGN KEY constraint failed
<rekado>after “guix gc” the repair worked fine
<rekado>and cuirass no longer barfs
<civodul>it should be impossible to export an invalid item
<apteryx_>Hello guix!
<rekado>but we’ve lost store items, of course.
<civodul>and thus impossible to copy it
<rekado>I have a copy of the store before gc, but not of the database
<rekado>I could recover it from the old server if needed
<civodul>it would be interesting to pick one of the broken store items (with an entry but no file, or vice versa), and see how we ended up with that
<rekado>I agree, but I had to be quick to restore cuirass on berlin.
<rekado>I’ll try later to replicate this on a separate machine.
<rekado>i.e. power up the old berlin head node and “guix copy” items to another node.
<civodul>well thanks for restoring the thing!
<civodul>that was the right approach :-)
<civodul>if you can replicate that's great, if you can't, we'll assume there's no bug ;-)
<rekado>I’ll try to make more time next year for working on cuirass.
<str1ngs>look like guix pull is fix with cc6a4b03867bc18baf30c0239c3c6dbf0709de7a
<civodul>str1ngs: awesome, thanks for testing
<bavier>uffda, atlas takes soo long to build... and then fails :(
<civodul>bavier: on aarch64?
<civodul>on x86_64 it takes ages but it succeeds
<bavier>on of the kernel benchmarks fails
<efraim>I should look back through the logs, I had a package that consistently failed on the firefly but succeeded on the odroid
<bavier>or maybe it was one of the tests. I lost the terminal output, and the logs are not helpful yet
<bavier>I might try again one shogun finished building with openblas
<bavier>s/one/once, cold fingers
<bavier>oh no, shogun needs texlive, I might have to actually fix it now
<rekado>from the output of cuirass: evaluate '(string-append binary-tarball. system)': 0.094 seconds
<mb[m]1>Is it okay to push a couple of heavy-impact patches to core-updates still? Stuff like TCL and krb5.
<mb[m]1>I see Hydra still haven't been able to evaluate it.
<mb[m]1>What's up with that, anyway.
<civodul>mb[m]1: i think it's ok
<civodul>i wonder what's up with Hydra
<civodul>it looks as if it evaluated it
<mb[m]1>Oh, didn't actually check recently.
<civodul>well the last eval still shows up as 12-09
<civodul>but the thing above said it was evaluated later
<mb[m]1>I tried starting one yesterday or so.
<adfeno>One of the issues I still don't understand how to find the true cause is the one with Emacs and GTK+.
<adfeno>Strangely enough, modifiying the Emacs recipe just so that it makes a dummy file works as expected (Emacs is built even with GTK+, and works regardless).
<mb[m]1>Heh, there's an empty icu60.2 directory on the ICU download page, and the front page has this news entry:
<mb[m]1>2017-12-12: ICU 60.2 maintenance update is not available. See ICU 60 download page for details.
<bavier>efraim: have you been able to build texlive-bin on your odroid and/or bbb?
<catonano>I built Polarri and I get this
<catonano>say that I want to know which packages depend on gjs, what's the command ?
<bavier>catonano: 'guix refresh -l <package>'
<bavier>catonano: which lists just the top-level dependents
<catonano>bavier: thanks
<quiliro>hello guix!
<bavier>hi quiliro
<quiliro>bavier: what's new?
<adfeno>Hi quiliro ;)
<bavier>quiliro: just trying to fix some more packages on my aarch64 system
<quiliro>adfeno: nice to see you too
<quiliro>bavier: what device?
<bavier>quiliro: it's a thunderx2 system
<quiliro>bavier: what is it good for?
<bavier>quiliro: hopefully for lots of number crunching
<quiliro>would be nice if there was pragmaos for newbies
<quiliro>bavier: fpga?
<quiliro>number crunchin=good for video?
<bavier>quiliro: I guess it would be, I'm interested in using it for HPC applications
<quiliro>such as?
<bavier>whatever the customers want :p
<quiliro>has anyone made an xorg.conf file on guixsd?
<bavier>quiliro: I made one the other day for my trackball mouse, but I don't have access to it now to share
<quiliro>bavier: i read now about hpc.....i still don't see how i could use it...maybe discovering how i myself think!
<quiliro>bavier: where is it supposed to go in guixsd?
<quiliro>bavier: would hpc be good for rendering blender?
<quiliro>found this for making my brightness on macbook air work
<quiliro>but i know /usr/share/X11/xorg.conf.d/10-nvidia-brightness.conf does not exist in GSD
<quiliro>so i suppose it should go in /etc/ ....right?
<efraim>bavier: I don't remember if texlive-bin built or not, checking it now
<efraim>Looks like it built
<quiliro>i cannot find X11 directory in /etc
<quiliro>i guess i should not touch /gnu/store/wa1fc13mnch9xb39v8r8jjbbpak4g7a9-xorg.conf.d
<adfeno>quiliro: HPC is short for "High Performance Computing", is basically a set of computers that are made available by someone... to someone else, and which are very fast than normal computers.
<adfeno>However HPC scenarios are known to limit the end-user (the person who bought the service).
<efraim>I wish I knew how to fix this macbook trackpad issue, it only goes up and down and switches between virtual desktops, have to reboot repeatedly or use an external mouse to get around it
<adfeno>quiliro: So the end-users who want to use Guix will have to use some hacks in order to use Guix (and packages from it) in HPC.
<efraim>finally got it working again
<quiliro>adfeno: thank you!
<quiliro>so hpc is good when using it on your own hardware
<quiliro>not as a service
<efraim>bavier: are you choosing which packages to build or are you using some form of script?
<quiliro>any message for me while i was out?
<adfeno>quiliro: No ;)
<bavier>efraim: I'm just wandering about in the package graph. Mostly trying to get (gnu packages maths) fully built
<adfeno>quiliro: Indeed, using it in your own HPC is very good (because HPCs are costly).
<bavier>efraim: thanks for checking
<adfeno>... but on the other hand Guix already has some documentation on how to use it on HPCs provided by other people.
<quiliro>found this
<quiliro>with this:
<quiliro>for configuring geforce 9400m
<quiliro>is this current?
<quiliro>so i could put the config detailed in
<quiliro>from /usr/share/X11/xorg.conf.d/10-nvidia-brightness.conf to /etc/xorg.conf.d/10-nvidia-brightness.conf
<jonsger[m]>ACTION works again on updating the openSUSE package of guix :)
<apteryx_>I'm getting this error:
<apteryx_>Seems to be the same class of problems that were triggered by a failing bayfront substitute server.
<bavier>apteryx_: have you tried '--fallback'?
<apteryx_>I haven't, good idea.
<apteryx_>It seems to be a valid workaround, although it is still suboptimal that it didn't try the next substitute server instead of hard crashing.
<civodul>apteryx: 504 "Gateway Timeout" means that is overloaded :-/
<civodul>so yeah, --fallback in the meantime
<rekado>but why is the *mirror* overloaded?
<rekado>that doesn’t seem right
<civodul>504 means that the mirror's backend didn't respond in a timely fashion
<civodul>so it's really hydra. that's overloaded, not mirror.
<rekado>oh, I see.
<quiliro>how can i know which config.scm i used? i have used several and i do not know which one i worked
<rekado>you cannot.
<quiliro>this is for 'guix system reconfigure config.scm'
<quiliro>i have '(file-system (device "/dev/sda1") (mount-point "/boot/efi") (type "vfat"))' on one configuration and:
<quiliro>'(file-system (device "/dev/sda1") (mount-point "/boot/efi") (type "msdos"))' on another
<quiliro>i have '(bootloader (grub-configuration (device "/dev/sda")))' on one configuration and '(bootloader (grub-configuration (grub grub-efi) (device "/dev/sda")))' on the other
<quiliro>i wonder which one is correct
<bavier>I see texlive-bin's luajiterr.test is disabled on mips64, I wonder whether we should do the same on aarch64
<g_bor>rekado: I'm checking now what builds on icedtea-8 and ant/java8, and it seems that java-hamcrest-1.3 does not build. This is somehow related to qbox.
<rekado>this just happened:
<bavier>wow, haven't seen that before
<bavier>is it reproducible?
<civodul>rekado: is it core-updates?
<rekado>no, that’s master
<civodul>the same command works for me on master, and berlin has substitutes
<civodul>not sure what's going on
<rekado>yeah, I’m running the command again
<rekado>it’s downloading all things again
<civodul>uh, scary
<civodul>does "dmesg" say something?
<rekado>ah, disk is full
<civodul>it's interesting that this makes Guile crash :-)
<civodul>that'd explain "Cannot exit gracefully when init is in progress; aborting."
<civodul>presumably Guile writes to stdout/stderr, which guix-daemon pipes to /var/log/guix/drvs
<civodul>but that fails, so maybe Guile gets EPIPE or something
<civodul>and since it's early on, it aborts
<civodul>and then something goes wrong
<rekado>cuirass on berlin doesn’t seem to be doing anything :(
<rekado>I implemented “guix offload status” which runs machine-load on all build nodes, and it shows me that they are all idle.
<civodul>it seems to have done something
<civodul>oh nice
<civodul>i can't strace cuirass because i can't sudo :-)
<rekado>I’ll send you a temp password in a moment
<quiliro>please tell me which is correct for my config.scm with uefi
<bavier>quiliro: presumable the one with 'grub-efi', but I do not have experience with efi
<quiliro>bavier: how about msdos versus vfat?
<quiliro>how can i know what type is my partition?
<civodul>quiliro: the file system type is "vfat", see the example at
<civodul>uh, on hydra, 'guix publish' is at 47M RES, and 2G VIRT
<noobly>Are there any use cases where Guix has strengths and NixOS has weaknesses, and vice versa?
<noobly>Sorry, *Nix, not NixOS
<noobly>A comparison of the languages is what I'm seeking
<quiliro>civodul: one of those two configurations (without 'bootloader-configuration') worked
<civodul>noobly: the rationale for using Guile Scheme is that it's a "complete" language, with libraries, compiler, debugger, etc.
<civodul>normally that makes development nicer than in an external DSL like the Nix language
<civodul>and also allows us to write better integrated applications, like UIs
<quiliro>what is the difference between:
<quiliro>(bootloader (bootloader-configuration
<quiliro> (bootloader grub-efi-bootloader)
<quiliro> (target "/boot/efi")))
<civodul>that said the Nix language is pretty fun to work with, being purely functional
<quiliro>(bootloader (grub-configuration (grub grub-efi)
<quiliro> (device "/dev/sda1")))
<civodul>the latter is deprecated and probably doesn't do what you want :-)
<noobly>civodul: thanks, I was browsing the haskell forums and a lot of them like Nix and say the language is great for their use cases, and I saw no mention of Guix so it made me curious
<civodul>well Guix came after Nix
<civodul>there's a couple of papers at that explain the rationale
<kmicu>noobly: that's strange, b/c any Haskell dev I know says Nix sucks, b/c it lacks a type system :) Guile/Guix at least has mature tools to help.
<catonano>should I want to see the code of the Gnome packages updater, where would I look ?
<catonano>ah guix upstream
<noobly>kmicu: well I thought perhaps the reason GuixSD is mentioned less is simply because it's currently less popular. I'm not a dev by any means and just interested in GuixSD and NixOS simply for their easily reproducible configs. Would you say Guix might be even better suited for Haskell development?
<noobly>and thanks civodul
<kmicu>noobly: Nix lang 'sucks' so badly that Haskellers created Dhall to generate Nix :)
<noobly>lol sounds cumbersome
<kmicu>Nix pain forces people to do such things. ;)
<noobly>Ok, I think I'm certainly going to work on transitioning towards GuixSD then. That was my original intention but just haven't gotten around to buying a blob free wifi chip for my x60
<bavier>catonano: guix/import/gnome.scm
<kmicu>(This is why I should write dhall-to-guile ;)
<catonano>bavier: thanks
<noobly>and i'm not good enough to roll my own nonfree pkgs
<civodul>rekado: cuirass is waiting, as if it thought it had nothing to do
<rekado>it’s frustrating that it doesn’t say much at all :)
<kmicu>noobly: Guix(SD) is libre, so you can modify every aspect of it and e.g. you can add blobs, but you are on your own with that. Search popular code hosting platforms for examples.
<kmicu>noobly: and do not forget too add entertaining comments like here
<Stream>how can i add persistent kernel parameters as in
<civodul>Stream: with 'kernel-arguments', see
<Stream>thanks also i got 2 monitors and there is a problem like the coursor icon is in one place but the click happens in another one, been trying to make mirrored monitors but while doing it i made it worse so now after i login it just doesnt load desktop and goes back to login screen but if i use other user that has mirrored already set up it works