IRC channel logs

2017-11-23.log

back to list of logs

<bms_>Hello.
<marusich>Hey there, Guix!
<rekado>Hi Guix!
<rekado>anyone here using yasnippet?
<rekado>I have a snippet for writing package definitions: http://paste.debian.net/plain/997186
<rekado>librecad does not show its icons any more.
<rekado>it accesses them via QIcon(":/icons/foo.svg")
<rekado>this means that the icons should be part of the resources that become part of the binary. My strace foo doesn’t help here.
<civodul>Hello Guix!
<roptat_>Hi :)
<roptat_>what happened to the php-fpm service patch? Is someone still working on this?
<civodul>good question
<civodul>looks like it fell through the cracks no?
<civodul>if you think it's ready, please go ahead and push it
<roptat_>I've lost some of my mails recently and it seems this patch is part of that... the archive modified the patch, the author is address@hidden
<roptat_>would you mind forwarding that mail to me?
<rekado>roptat_: are you not using debbugs?
<nee`>I'll finish it soon, sorry didn't have time for it recently
<rekado>it’s bug 28769
<roptat_>oh, of course
<roptat_>nee`: is the issue with the log-file fixed?
<nee`>I remember I got a reply, but I didn't get around to look at the system test in it.
<nee`>Yeah, I remember that everything was fixed. I just need to look do the test and maybe discuss some naming.
<roptat_>ok, great :)
<rekado>ACTION builds the disk image now
<rekado>Renjin looks interesting.
<rekado>It’s an implementation of R on top of the JVM.
<rekado>to achieve this they wrote a compiler that binds to GCC and builds Java classes
<rekado>(is this derived from GCJ?)
<rekado>would take a lot of effort to package this. Lots of dependencies.
<brendyn>I'm trying to build pcmanfm's debugging symbols by adding a debug output, but then when i run gbd pcmanfm I get CRC mismatch error
<civodul>rekado: there's also a research group at Northeaster that develops an R implementation on top of LLVM, IIRC
<m-o>Hi guys, just hit the guix memory overflow issue on an ARM machine with 2G of RAM on latest guix screenshot (953c2de) : GUILEC gnu/packages/unrtf.go
<m-o> GUILEC GC Warning: Failed to expand heap by 8388608 bytes
<m-o>GC Warning: Failed to expand heap by 8388608 bytes
<m-o>
<civodul>s/guys/people/ :-)
<civodul>m-o: did it crash?
<civodul>what you're showing here is a warning
<civodul>it likely comes from a weak table issue that has been fixed in Guile
<m-o>Oops sorry work habit :(
<civodul>rekado: i think it's this one: https://github.com/duncantl/RLLVMCompile
<m-o>Nope then it crashed
<civodul>OOM?
<m-o>Too many heap sections: Increase MAXHINCR or MAX_HEAP_SECTS
<m-o>/gnu/store/dl7fwabzl32df7apnbirrsq2lchm0y38-bash-minimal-4.4.12/bin/bash: line 7: 11885 Aborted
<m-o>port/pypi.scm guix/import/stackage.scm guix/scripts/import/crate.scm guix/scripts/import/gem.scm guix/scripts/import/json.scm guix/scripts/import/pypi.scm guix/scripts/import/stackage.scm guix/ssh.scm guix/scripts/copy.scm guix/store/ssh.scm guix/scripts/offload.scm guix/config.scm guix/tests.scm guix/tests/http.scm
<m-o>make[2]: *** [Makefile:5300: make-go] Error 134
<m-o>make[2]: Leaving directory '/tmp/guix-build-guix-0.13.0-11.953c2de.drv-0/source'
<m-o>make[1]: *** [Makefile:4431: all-recursive] Error 1
<m-o>make[1]: Leaving directory '/tmp/guix-build-guix-0.13.0-11.953c2de.drv-0/source'
<m-o>make: *** [Makefile:3021: all] Error 2
<m-o>phase `build' failed after 4469.3 seconds
<m-o>note: keeping build directory `/tmp/guix-build-guix-0.13.0-11.953c2de.drv-0'
<m-o>builder for `/gnu/store/fdgb71yl808n6ia1ccvhdl3wim6qa0hx-guix-0.13.0-11.953c2de.drv' failed with exit code 1
<civodul>enough :-)
<m-o>guix system: error: build failed: build of `/gnu/store/fdgb71yl808n6ia1ccvhdl3wim6qa0hx-guix-0.13.0-11.953c2de.drv' failed
<m-o>
<civodul>please use a paste bin
<civodul>can you try reducing the number of parallel jobs?
<m-o>Mhh this is a guix build triggered on a "guix system ..." command
<m-o>so i don't know if it is possible\\
<brendyn>civodul: I just read your reply about nautilus/pcmanfm. i build pcmanfm with debugging symbols now im wondering how to do what you suggested, as i have no experience with gdb
<civodul>m-o: try --cores=1 for instance
<m-o>ok i'll try thanks :)
<civodul>new article! https://guix-hpc.bordeaux.inria.fr/blog/2017/11/installing-guix-on-a-cluster/
<civodul>:-)
<rekado>woo!
<rekado>civodul: in the summary it says “mater node” but should say “master node”
<rekado>or maybe you *did* mean mother node…
<civodul>arf, indeed
<efraim>Got my internet working again, turns out the internet company blacklisted the MAC address of my router, probably a money grab, once I told it to use one from another machine on my network it suddenly started working again
<civodul>uh, terrible
<efraim>Also got an email ad from digital ocean, 3GB of RAM, 20GB SSD, $15/month
<efraim>'Plenty of RAM for guix pull!' Now to make a service for quassel and ttrss
<rekado>civodul: boss already tweeted it: https://twitter.com/AltunaAkalin/status/933709977390379010
<rekado>testing the disk image now
<rekado>booted fine
<rekado>the number of packages in the manual is wrong. It’s no longer just 5,300 packages.
<civodul>good point
<brendyn>I found a TODO for adding an input from two years ago that could have been done nearly 2 years ago
<civodul>rekado: re tweeter, it's nice to have a boss who really cares!
<civodul>brendyn: heh, that happens :-)
<brendyn>It also happened to fix I problem I had, and I found it accidently
<brendyn>also a CVE patch.
<efraim>I think aarch64 is almost at 6100 packages
<civodul>efraim: pretty good!
<rekado>the manual still recommends starting with the desktop configuration, not with bare-bones.
<rekado>should we change that?
<efraim>I should go back to working on Java
<efraim>Orthogonally, we should change the desktop configuration to comment out both GNOME and xfce and instruct users to uncomment the one they want
<efraim>Many of the requests for help include GNOME removed from the desktop config
<civodul>rekado: you mean because of missing substitutes and things like that?
<rekado>yes
<rekado>The section “Proceeding with the Installation” has a check list for the config file.
<rekado>One thing that’s missing in my use cases is to check for kernel modules.
<rekado>such as sata_nv
<jonsger>civodul: nice blog post :) I didn't know that package search and browser on GuixHPC
<efraim>Also whatever raid driver is needed for btrfs raid
<brendyn>Has anyone setup their emacs to automatically input template commit messages? I find myself tying the same thing over and over.
<civodul>jonsger: thanks, roelj implemented it
<civodul>rekado: right
<jonsger>maybe you or roelj could send an email to guix-devel to inform the community... :)
<rekado>brendyn: I have.
<rekado>I have a yasnippet for it, and previously used a macro.
<civodul>jonsger: agreed, i think roelj is too shy, i'll ping them :-)
<rekado>Should I add my snippets to the repository’s etc directory?
<efraim>I think there's a git option also
<rekado>installing GuixSD on this server is very slow.
<rekado>it’s building Guix now, but … it’s weird
<efraim>Something about git commit.template
<rekado>it prints a couple dozens of lines of “GUILEC …”, then freezes for minutes, and then repeats
<rekado>efraim: right. But it isn’t context-sensitive.
<efraim>Right
<rekado>my snippet can call out to magit and get the list of changed files and arrange them
<rekado>but a template could be useful for the <1% of people who don't use Emacs ;)
<rekado>ACTION opens up vi
<rekado>ACTION exits vi
<vagrantc>opening and exiting is on the path to mastery
<rekado>civodul: installing GuixSD I get a lot of messages about deprecated features being used.
<rekado>the installation is very noisy even without them, but seeing these messages repeatedly can be confusing to new users.
<rekado>also: there are a few more conflicts: util-linux and shadow both provide “sbin/nologin”
<rekado>and inetutils and net-tools both provide “ifconfig”.
<efraim>I think we're depreciating net-toold
<efraim>s/toold/tools
<rekado>it still seems to be part of the standard installation.
<rekado>I only used a bare-bones config with static networking
<rekado>okay, installation of GuixSD with the current disk image is complete. The system boots and there was much rejoicing.
<brendyn>rekado: I'm intersted in them. I use Emacs but I'm elisp-impaired.
<civodul>cool
<civodul>rekado: i'm supper annoyed about the Guile deprecation warnings
<civodul>they all come from the use of _IONBF et al.
<civodul>i agree we should remove them, but it's not necessarily easy
<civodul>problem is we need to set GUILE_WARN_DEPRECATED=no in some derivations, but we can't do that in all the module-import derivations because that'd lead to a full rebuild
<civodul>rekado: maybe this would work: https://paste.debian.net/997264/
<civodul>given that computed-file is used almost exclusively in GuixSD
<brendyn>odd. nautilus is wrapped in a script that does nothing but launch it.
<CharlieBrown>Emacs isn't seeing my Guix-installed font on a foreign distro.
<efraim>You might need to update your fontcache
***pksadiq_ is now known as pksadiq
<kmicu>"vi /etc/systemd/system/guix-daemon.service", but you did not write how to quit vi :/ ;p
<janneke>*woot* my tcc-boot package built!
<CharlieBrown>kmicu: You may want to ttry mi vimrc: https://notabug.org/csh/dotfiles/src/master/.vimrc
<numerobis>haha!
<mb[m]1>Re: guile warnings, there is still time for full-rebuild changes on core-updates.
<DusXMT>filtered-port: failed to execute ' -dc ': No such file or directory
<DusXMT>guix substitute: error: corrupt input while restoring '/gnu/store/xa391b23r5lbwxb9q26sq5rq1fkd1xi3-glibc-utf8-locales-2.25' from #{read pipe}#
<DusXMT>Anyone else experiencing this?
<DusXMT>(when trying to update or install anything)
<DusXMT>dc seems to be in the path, available both to the daemon and the user
<vagrantc>janneke: congrats!
<janneke>vagrantc: thanks!
<janneke>here's a dependency diagram in svg: http://paste.debian.net/997274/
<marusich>DusXMT, I haven't seen that, myself, recently. Without further information, my best guess is that perhaps the connection between your machine and the substitute server was interrupted for some reason. For now, you can probably work around it by using --fallback.
<marusich>It is also possible that a substitute is corrupt on the substitute server, but without further information, I'm inclined to suspect your connection first.
<marusich>rekado, I for one would appreciate your snippet for Emacs
<efraim>CharlieBrown: your Tilda works? Mine insists on making a new config each time I launch it.
<nee`>roptat_: I sent an update for the php-fpm patch https://debbugs.gnu.org/cgi/bugreport.cgi?bug=28769#29
<mwette>Why does the guix daemon (need to) run as root?
<mb[m]1>Any suggestions for how to organize patches from the glibc stable branch? I'm thinking just using numbers in the order they were added. Categorizing them all will be a lot of work, and not all are connected to a bug/have sufficient context.
<civodul>mwette: for process isolation; see https://guix-hpc.bordeaux.inria.fr/blog/2017/09/reproducibility-and-root-privileges/
<mwette>Ah. The builds are run as the client user. And it can be run as non-root-user provided --disable-chroot is provided.
<civodul>yeah but using --disable-chroot is not a great idea
<civodul>mb[m]1: you want to add more patches than the 3 you posted?
<mb[m]1>civodul: See Leos reply. I only picked a small subset, but there are many other important-looking ones.
<mwette>civodul: Thanks for the reference.
<civodul>mb[m]1: oh i haven't seen Leo's reply yet
<civodul>it's a bad situation, we end up doing upstream's work
<mb[m]1>I'm going to bump the discussion about stable releases on libc-alpha as well, silly that every (rolling) distribution is supposed to comb through all ~100 patches.
<civodul>silly and risky, too
<civodul>we're not in a position to judge the pros and cons of 100 patches
<mb[m]1>True.
<rekado>DusXMT: “failed to execute ' -dc '” — that’s not the full command
<rekado>“-dc” is just an argument to … gzip, I think
<rekado>so this indicates that you have an unconfigured Guix.
<rekado>are you using Guix from a git checkout?
<rekado>marusich: I’m going to send a patch to guix-devel
<rekado>they are not perfect – the snippet language is very limited, but they are good enough for my purposes.
<vagrantc>would it make sense for guix system init/reconfigure/etc. to store the foo.scm used to create it?
<rekado>vagrantc: there’s a system service for that!
<vagrantc>aha!
<rekado>now if only I remembered what it was called…
<vagrantc>ACTION is still fairly lost, but having fun stumbling around
<rekado>looks like it was never actually added to Guix proper.
<rekado>Message-ID: <871snzfl4s.fsf@gnu.org> shows how to do it
<rekado>here’s the relevant part: http://paste.debian.net/997311/
<mb[m]1>TIL git has a "notes" subcommand: https://git-scm.com/docs/git-notes