IRC channel logs

2018-04-30.log

back to list of logs

<davidl>snape: /var/lib/certbot/renew-certificates - https://paste.debian.net/1022611
<davidl>snape: with nginx user I can access /etc/letsencrypt/{csr,renewal,renewal-hooks} but not &/{accounts,keys}.
<noonian>Hello, I just installed GuixSD on my laptop successfully but after running `guix pull` I cannot reconfigure the system. `guix reconfigure /etc/config.scm` complains about not having "nvme" in initrd-modules (I gather this is required to detect my ssd drive). When I add nvme to the initrd-modules reconfigure proceeds but fails with `module not found "nvme.ko"`. Any help would be appreciated :)
<buenouanq>How are you declaring it? post the config
<noonian>buenouanq: http://paste.debian.net/hidden/e3b86201/
<noonian>I found this irc log mentioning the missing module https://gnunet.org/bot/log/guix/2018-01-29 and this mailing list thread discussing changes with nvme https://lists.gnu.org/archive/html/guix-devel/2018-01/msg00492.html but I'm not sure what they imply for using guix with nvme drives
<mange>Maybe try leaving nvme out of your initrd-modules list and run reconfigure with --skip-checks. I don't quite understand the issues involved, but I have had the module autodetection incorrectly tell me that I've needed to add things to that list.
<mange>Worst case, you can just boot back into the working generation and try something else.
<noonian>mange: I think that worked! Thanks!
<efraim>I got the setuid bits working for enlightenment, next up is the screen locker, then the lanuage switcher
<snape>sneek: later tell davidl: it seems like server1.selfhosted.xyz/.well-known returns 404 instead of 403. That's a hint. Could you please share your nginx.conf?
<sneek>Got it.
<divansantana_>civodul: so my grub issue was due to me trying to use mbr on uefi system. Even though the BIOS it was set to both. Think it's buggy.
<snape>(civodul isn't here)
<divansantana_>snape: yeah, I know but he would still get the msg later wouldn't he?
<divansantana_>Why does luks encrypted system ask for the passphrase twice? Once for grub to read boot data and once to decrypt root volume. Is this the case for everyone?
<divansantana_>Also wondering how to set both keyboard layouts to dvorak. Will look into that just now.
<zybell_>snape:.well-known/ too? note the slash! Without it it supposed to either 404 or 30.!
<buenouanq>I've had that same problem with MBR and forgot divansantana_
<buenouanq>fewest headaches for me have been using GPT and the efi-grub
<buenouanq>What do you mean by both keyboard layouts?
<buenouanq>I can help if you're referring to making slim dvorak.
<zybell_>buenouanq:obviously in grub and in linux. passwords and keyboard layout!
<buenouanq>the statement before the last didn't register
<ngz>Out of curiosity. Is there a way to create a package from a software that requires npm in the first place for its installation?
<efraim>I'm not sure how much work has been done with npm so far
<ngz>I was wondering if there was a way to handle the package at a more lower level
<ngz>I.e., find the dependencies, build them...
<efraim>Like the recursive importer?
<ngz>Yes
***mr-jonze is now known as bill-auger
<efraim>Might have to search the maillinglist for it, not sure if it was ever committed
<efraim>rekado: do you know? ^
<snape>divansantana_: if he checks the history, yes, but he might not
<snape>you can use the bot (sneek) to make sure he gets notified
<snape>sneek: later tell divansantana_: test
<sneek>Got it.
<snape>zybell_: I don't know about the slash, but it returns the same either way
<divansantana_>snape: ok cool. Good to know. I figured most have something like znc setup. But suppose then they would show as online or away. Or some other mechanism.
<sneek>divansantana_, you have 1 message.
<sneek>divansantana_, snape says: test
<divansantana_>snape: swt
<snape>divansantana_ but I'm pretty sure civodul checks https://gnunet.org/bot/log/guix
<divansantana_>snape: yeah thats cool and useful
<zybell_>from https://github.com/gg-anon/gg and http://stanford.edu/~sadjad/gg-paper.pdf I got the idea that you can run any build under `strace -o log -f` and convert the resulting log file to a description of the dependencies,the needed programs...short to a .scm file.
<zybell_>ngz efraim ^
<civodul>Hello Guix!
<efraim>Hi!
<g_bor>hello guix!
<sneek>Welcome back g_bor, you have 1 message.
<sneek>g_bor, rekado says: I haven’t sent out the third email yet.
<g_bor>I have a few question.
<kmicu>( ^_^)/
<g_bor>1. I have seen communication with uniq10, ans sahithi. Is there a repository where we can keep track of their work?
<g_bor>2. I have not seen communication from the par of the other two GSoC interns. Have I missed something?
<roptat>hi guix!
<roptat>it seems neither hydra nor berlin has IPv6. Is it a limitation of the network they're on, or a missing configuration?
<divansantana_> /etc/guix/machines.scm , how does one configure this in guixsd? Is it correct to place the file "manually" in /etc/guix or does one specify this in the main config.scm file?
<snape>divansantana_: you just need to edit it manually
<snape>Guix knows about it, so no need to talk about it in config.scm
<divansantana_>snape: cool. Thanks.
<ArneBab_work>janneke: you’ll want to update the mes status note in http://bootstrappable.org/projects.html with your most recent progress report http://lists.gnu.org/archive/html/guile-user/2018-04/msg00033.html
<janneke>ArneBab_work: thanks for the headsup and thanks for your kind words!
<roelj>What happened to the build of ‘guix’ at hydra.gnu.org/build/2654230? I cannot reproduce it with ‘guix build guix’.
<zybell_>roelj:you didnt run out of diskspace?;-) Its 100 a runit exit code.
<civodul>roelj: it's the dreaded non-deterministic crash :-/
<civodul>there's a couple of bug reports about this
<civodul>i hope to squash it at some point
<civodul>success here: https://berlin.guixsd.org/log/wpmcqyn3spv8ljmqs4b8j0kh8vcjv4bw-guix-0.14.0-11.ab85cf7
<civodul>i've restarted the build on hydra.gnu.org
<ngz>Hmm. How can I have Guix-installed software appear in Gnome Activities menu? I had this, re-installed my system and...
<civodul>i think it happens automagically no?
<ngz>Apparently not.
<ngz>May I have to mess with .gnomerc ?
<civodul>which applications are missing? those installed in ~/.guix-profile?
<ngz>Yes
<civodul>i don't use GNOME, but on another compute with GuixSD/GNOME, it works fine
<civodul>applications are not automatically added to the dock but you can search for them and add them
<civodul>it took ~25mn between the time i pushed fb8966abc and the time berlin.guixsd.org was done building guix-modular on all 4 arches
<efraim>It looks like my screen lock problem in enlightenment may have just been a 0.22.2 bug
<divansantana_>With offloading, Can one add localhost to the machines.scm file so as to speed up compiling as the compiling/load/work is spread in parallel between the remote host and localhost?
<roptat>g_bor: guix-manual is on the TP \\o/
<roptat>I remember you were interested in translating it
<roelj>civodul: Thanks for restarting the build on Hydra. Where can I see it?
<pkill9>is the KDE desktop environment available on Guix?
<ngz>pkill9: I think it's still missing a few packages
<civodul>roptat: woohoo, thank you!
<civodul>roelj: same URL as before
<civodul>efraim: the freeimage issue with guile-sly is still there, do you have any idea?
<g_bor>roptat: thanks! I will check it out.
<civodul>roptat: which reminds me, we'll have to think about the URL scheme and scripts to publish the manual at gnu.org/s/guix/manual
<civodul>the "gendocs.sh" script in Gnulib is monolingual
<roptat>civodul: can we trick it to think there is more than one manual it has to build
<roptat>?
<civodul>we'd have to run it several times i guess
<civodul>but it's really low-tech, so we could just as well come up with our own scripts
<janneke>ACTION has just updated wip-bootstrap for mes 0.13 and co
<civodul>janneke: woohoo, sounds exciting!
<efraim>civodul: I can check the build logs later and check it against other errors
<janneke>civodul: thanks...yes, the first `tcc-boot' that has been built without gcc,glibc or guile
<janneke>other than the `package-with-bootstrap-guile'-guile to run guix...
<civodul>janneke: so 'tcc-boot' is built with a (pre-compiled) Mes + MesCC?
<janneke>yes
<janneke>the *-boot packages all still have %*-seed counterparts
<janneke>for mes-boot, that's %mes-seed, which contains mes.M1 (assembly-like compiled output from a previous MesCC)
<janneke>so two directions for me/mes now: up and down
<janneke>i'm working with OriansJ to replace this mes.M1 assembly-like seed with a mes.M2 seed that's essentially simplified and preprocessed C code
<janneke>that was down, for up: mes-tcc passes 67/68 tests...but fails to compile itself
<efraim>That seems like the most important one :)
<janneke>it looks like we're close
<civodul>woow, impressive
<civodul>could you send a status updated on wip-bootstrap to the list?
<janneke>mes-tcc segfaults when just before dumping ELF for tcc
<civodul>with all the nitty-gritty details :-)
<civodul>ok
<janneke>civodul: yes, will do; wanted to update wip-bootstrap first
<civodul>yes sure
<civodul>actually that could go to guix-blog@gnu.org as well, muahaha
<janneke>ah, yes -- hmm
<civodul>:-)
<davexunit>has anyone ever seen an error like this? 'guix environment: error: corrupt input while restoring archive from #<closed: file 79f7d90>'
<davexunit>I seem to have a degraded guix installation on an ubuntu workstation I use at $dayjob
<davexunit>or 'guix package: error: corrupt input while restoring archive from socket'
<davexunit>I get some error like this whenever I try do use something that interacts with the daemon :(
<mbakke>Hmm, this commit broke our meson patch; not sure what to make of the commit message: https://github.com/mesonbuild/meson/commit/7c37ca15f38dce30eeef6cd8bc53f25406f9029e
<thomassgn>playing around with containers and think I don't understand some things here. So I run 'guix system container id.scm --root=id-root' and then 'sudo ./id-root --expose=shared' to run it. Then I use 'sudo guix container exec $(pgrep id-root) /run/current-sy.../bin/bash --login' to get a shell. In this shell, supposedly in the container, I get hostname returning my hosts hostname, not the one set for the
<thomassgn>container, it also looks like the containers root is my hosts root, and I have write access to ./shared ... wat?
<dustyweb>hi hi
<dustyweb>which package provides all the "common" fonts, such as monospace, dingbats, etc?
<dustyweb>I am having a number of issues, which may be due to hard drive corruption (I accidentally shocked my computer pretty badly) while setting up this new laptop. one of the issues is a lot of fonts aren't showing up, but i think these ones are pretty common if you have xorg installed I thought
<dustyweb>unfortunately since I'm on the road, diagnosing these things is hard ;P
<davexunit>thomassgn: perhaps it's not actually in the container?
<dustyweb>also hi davexunit, we're in a moving van headed closeish to you :)
<davexunit>I wrote this back in 2015 but it was a proof-of-concept that I didn't know people were actually using ;)
<davexunit>dustyweb: oh wow! en route!
<dustyweb>davexunit: yes but it's not the full move
<davexunit>dustyweb: unfortunately I'm not sure what the font package is you are looking for
<dustyweb>I am now regretting not having set up this laptop with btrfs now
<davexunit>and the guix setup on the machine I'm on is borked
<dustyweb>maybe I would be having an easier time telling if I was hitting corruption issues
<dustyweb>bork bork bork
<thomassgn>davexunit: yes, that's what it looks like.
<davexunit>thomassgn: in that case: my sincerest apologies!
<davexunit>though I would have expected an error if it failed to join a namespace
<thomassgn>hehe, no worries. namespace?
<davexunit>that's what the underlying kernel technology is called
<thomassgn>right
<davexunit>the container creates many namespaces, each isolating the process from some global kernel resource
<thomassgn>maybe that's why I have to run guix container with root?
<davexunit>part of it, yeah
<thomassgn>cause without root/sudo I don't get access
<thomassgn>just a failed with status 1
<davexunit>all namespace operations require superuser
<davexunit>unless you are also creating a user namespace (and your OS grants unprivileged users access to user namespaces)
<davexunit>in practice that often means needing to be root
<thomassgn>right. ok
<davexunit>the 'guix container exec' command is supposed to look up all the namespaces for the given pid, and then join those namespaces, then run the command you specify
<davexunit>so definitely ensure that the pid is correct
<davexunit>dustyweb: I don't see any package that aggregates a ton of fonts, but font-dejavu is a common font package
<davexunit>I don't know if that helps in the slightest
<thomassgn>how would I do that? I mean I get one pid from pgrep for the launch script/root. I could pgrep for qemu or something else though. Or is there some other better way of finding a pid?
<davexunit>you need the pid of the process started by 'guix system container'
<davexunit>since this is just proof of concept we have no convenient tooling to deal with these things. you just have to look it up in the process table.
<efraim>I also like terminus
<efraim>As a font
<thomassgn>davexunit: Thanks, I'll look more into it, the possibilities are very interesting I think. But maybe a VM is easier for now.
<davexunit>yeah a VM is a more well-supported path
<davexunit>the container stuff is an interesting experiment that needs more work
<rekado>ngz, efraim We have a recursive cran/bioconductor importer. AFAIK this hasn’t been generalized to other importers yet.
<rekado>“man avr-gcc” shows me “timestamp”
<davexunit>ah, software
<davexunit>ACTION throws computer into garbage
<bavier`>haha
<ngz>rekado: OK, thank you for the info.
<civodul>sneek: later tell dustyweb for common fonts, see the 'gs-fonts' package
<sneek>Okay.
<lfam>Missing fonts was the problem that helped Mark find the obscured store references that silently broke grafting and gc.
<civodul>yeah
<lfam>I wonder if dustyweb recently ran gc
<civodul>lfam: though i think (hope?) this specific bug is gone
<lfam>Yes, hope so :)
<civodul>roptat: since you're in Android ;-), could you take a look at https://bugs.gnu.org/31265 ?
<OriansJ>sneek: later tell janneke adding octal and binary really makes M1-Macro live up to the name mescc-tools
<sneek>Got it.
<OriansJ>sneek: botsnack!
<sneek>:)
<uniq10>Hi reepca. How do I build a package using the guile daemon?