IRC channel logs

2020-03-26.log

back to list of logs

<Veera>Hi guix
<Veera>How to know that a patch got accepted?
<Veera>Will it come from debbugs@gnu.org
<kondor[m]>Veera: you get a confirmation that your bug was closed
<kondor[m]>i got mine from help-debbugs@gnu.org :)
<Veera>oh
<Veera>kondor: irc help in kondor[m] does the [m] represents irc moderator
<kondor[m]>afaik it means that the user is using matrix to connect to IRC
<kondor[m]>matrix.org
<Veera>oh
***afk11` is now known as afk11
<vagrantc>hrm.
***ub_ is now known as ub
<brendyyn>There was a NixCon 2019 talk about Guix. They said many positive things, but also said guile's error reporting for syntax errors was a pain point.
<vagrantc>yeah, it usually just explodes in a bunch of random seemingly unrelated things being undefined
<brendyyn>If I understood better how guile worked id try to fix it.
<vagrantc>presuming i didn't botch it, there should be a very rudimentary supported kernel pinebook pro on master now :)
<Gamma02>Hello, Is there any optimization tips for "guix pull" ?
<Gamma02>Is there something similar to refresh mirror like in Arch ?
<roptat>Gamma02, not sure what you mean, but guix pull should be able to download substitutes (if they are available)
<roptat>I usually look at https://ci.guix.gnu.org, the modular-master job will give you some indication on what is available for your architecture on the ci, then you can use guix pull --commit=some-commit-number for a commit that has been built on the build farm
<hpfr[m]>I took a look at issues.guix.gnu.org. It’s very cool, but it seems like a bit of a less happy marriage between the email and web workflows than what sr.ht is doing. Has Guix ever looked into sr.ht as a means of better exposing the project to devs who are used to the web workflow? Just curious, thanks
***samis is now known as CompanionCube
<kiwi_44>HI
<kiwi_44> Anyone from outreachy?
***d4ryus2 is now known as d4ryus
<guix-vit`>Hi there.
*janneke aaarrrrgh, my emacs-pinentry went gui again
<janneke>Good morning, GUix!
<Blackbeard>@freenode_guix-vit`:matrix.org: hi
<Blackbeard>janneke: good morning :)
*janneke prepares patch for make-bootstrap on guile-3
<janneke>hi Blackbeard :)
<Blackbeard>janneke: that's awesome
<rekado>hpfr[m]: as the author of the software behind issues.guix.gnu.org I agree with you that it’s far from perfect.
<rekado>hpfr[m]: however, we are currently committed to using debbugs; issues.guix.gnu.org is just a fancier frontend to debbugs.
<rekado>if I had more time to hack I’d add support to commenting and submitting bugs via the web interface
<rekado>but I currently don’t have the resources to hack on that part.
<nataraj_>Anybody using Guix on Slackware? I have Slackware and Voidlinux on different partitions.
<nataraj_>Guix package manager works like a song on Void
<nataraj_>Aint working on Slackware
<nataraj_>guix (GNU Guix) 1.0.1
<rekado>nataraj_: how does it not work?
<nataraj_>rekado: http://sprunge.us/vs3P1K
<nataraj_> http://sprunge.us/nyBiAq
<rekado>so the daemon is fine but Guile segfaults?
<rekado>can you identify which process actually segfaults?
<nataraj_>yep, so, it could be guile?
<nataraj_>guile (GNU Guile) 2.2.6
<rekado>is this with the Guile package from Slackware or anything installed via Guix?
<nataraj_>ought to be from slackware package. Dont remember having installed separate
<rekado>nataraj_: maybe report this to the Slackware maintainers?
<rekado>you could also try to run Guile with gdb to see where it segfaults
<rekado>Guile segfaulting is a very fundamental problem that is unlikely to be caused by Guix.
<nataraj_>let me check. Which Guile version is the best for Guix?
<rekado>the latest version uses Guile 3.0
<rekado>but the most recent release was made with 2.2
<nataraj_>rekado, something strange while checking out on chroot on Voidlinux partition
<nataraj_>Guile aint installed in Void. But guix works. Bummer!!!
<usney>how can I refresh my list of apps I installed so I can run them? I can run them after I install them I just have to reboot. I will look at the guix reference card for answers while I wait for a response.
<nataraj_>I mean guile not in the path
<Blackbeard>usney: I am not sure what do you want to do?
<Blackbeard>You mean you install them but they won't start right away?
<Blackbeard>If so, are you using guix as a package manager or guix system ?
<usney>I am using guix as a package manager
<Blackbeard>usney: what WM or DE are you using?
<usney>I installed via the script and I have to do a little setup with the documentation afterwards
<usney>I am using pureos
<usney>have/had*
<Blackbeard>And what are you using to launch the guix applications
<Blackbeard>With a terminal?
<Blackbeard>If you are in pureOS that means GNOME
<Blackbeard>Are you trying to launch the guix applications using the GNOME launcher, I don't know what it is called
<DamienCassou>hi
<rekado>nataraj_: looking at “which guix”, what’s the shebang on Slackware?
<DamienCassou>Blackbeard: I've successfully used your approach to store /gnu/guix outside of my / partition. Thank you
<Blackbeard>Damien Cassou: I am glad it worked
<Blackbeard>usney: are you using GNOME launcher or a terminal to launch the applications
<DamienCassou>I have a problem with `./pre-inst-env guix system build some-config.scm` which displays a bunch of "source file is newer than compiled files" messages and then exits with status code 1 without saying anything else. This is in a Guix System VM.
<Blackbeard>Damien Cassou: did you do a git pull?
<DamienCassou>I did with both the main user and using sudo. The guix channel points to a git repo
<DamienCassou>a local git repo
<Blackbeard>Damien Cassou: did you run make after the git pull
<DamienCassou>the git repo is shared between the host (Fedora) and the qemu guest (Guix System). I ran bootstrap/configure/make from the host
<usney>they were cli apps so I was using the terminal Blackbeard
<usney>I don't think I have this issue for gui apps
<rekado>usney: you don’t need to reboot for packages to become available.
<rekado>usney: for terminal applications have you tried opening a new terminal?
<rekado>this causes your shell initailization files to be read, which should be configured to set up variables needed for using Guix software.
<Blackbeard>usney: interesting. Can you do "echo $PATH"
<rekado>you can also do this manually, and Guix usually prints instructions after installing something new.
<rekado>usney: in your ~/.bash_profile you should have lines like this: export GUIX_PROFILE=$HOME/.guix-profile; source $GUIX_PROFILE/etc/profile
<rekado>this will cause new shell sessions to do the right thing.
<rekado>note that the GUIX_PROFILE variable is important. Not setting it results in sourcing just one particular generation of your profile, which will not change as you install or remove software.
<usney>no I did not try opening a new terminal
<usney>I did the "echo $PATH" command and it shows I have the regular paths including the guix paths
<usney>thank you rekado
<usney>I do not have that file located anywhere "~/.bash_profile" rekado I did a updatedb and locate and couldn't find it.
<usney>Before I try and create that I am going to try your close and open new terminal idea and see if that works
<usney>I did cat ~/.bash_profile and said there was no such file then I used locate to see if it was somewhere else.
<usney>I do have a bashrc file maybe that is it?
<usney>I'll see what it says
<rekado>usney: do not use bashrc for this.
<rekado>usney: Bash treats these two files slightly differently
<rekado>~/.bashrc is used for *every* new shell
<rekado>~/.bash_profile is considered only for every so-called login shell.
<rekado>so whenever you run “bash” to create a sub-shell ~/.bashrc will be evaluated
<usney>cool thank you rekado
<rekado>if you put the above lines into ~/.bashrc you would break “guix environment”, because that spawns a sub-shell.
<usney>I found an example bash_profile in my "/" I will see if it helps any with making my own.
<rekado>it spawns a shell with new environment variables set … and then Bash overwrites all of these variables because of what you have in ~/.bashrc.
<rekado>so ~/.bashrc should only contain things that really should be set for every shell
<usney>I did not edit it I used cat to see it
<rekado>and ~/.bash_profile should contain things that should only be set once per session
<rekado>usney: I understand. I’m just explaining a scenario that has happened to a lot of users in the past.
<usney>so does that mean it gets overwritten?
<usney>thank you for your advice I appreciate you being so helpful
<rekado>you’re welcome!
<usney>so does the ~/.bash_profile get overwritten each time them? rekado
<usney>them/then*
<rekado>the *file* will not be overwritten, no
<rekado>it’s a configuration file under your control.
<rekado>I suggest adding the above lines once. (Create the file if it doesn’t exist.)
<rekado>the above lines cause the shell to set a bunch of environment variables
<rekado>these variables and their values are generated by Guix whenever you install software
<usney>should I add those lines with the example file in pureos?
<rekado>each Guix “profile” (e.g. ~/.guix-profile) contains a generated file with variables that are necessary to use the installed software.
<rekado>what example file?
<usney>let me pastebin
<rekado>ok
<usney> https://bpaste.net/raw/Z3GA rekado
<rekado>usney: you can add my suggested addition to the bottom of that file.
<usney>cool
<usney>So I can safely use that example file
<usney>it wasn't installed by default it was a system example
<rekado>oh
<usney>updatedb is so much faster I have notice I remember in the old days it would take a long time to update
<rekado>you can also just override it completely
<rekado>if you haven’t used it before
<rekado>a file with just the lines I suggested would be just fine.
<usney>okay cool will do
<usney>"echo export GUIX_PROFILE=$HOME/.guix-profile; source $GUIX_PROFILE/etc/profile > ~/.bash_profile"
<rekado>better use an editor for this
<usney>is there a permissions examining tool? I used mac os in the past long time ago it had one but I think it wouldn't list recommendations it would just do it automatically.
<usney>okay cool
<rekado>or quote everything after “echo” with single quotes
<usney>I like nano
<rekado>nano is fine
<usney>is emacs hard to learn?
<rekado>usney: what do you mean by “permissions examining”?
<rekado>usney: as a long time Emacs user (who is writing this from Emacs) I have to say: yes.
<usney>to make sure permissions are correct and not exposing your machine in a bad way.
<rekado>usney: Emacs has some odd defaults that can be confusing.
<usney>cool I got a ereader recently so I can easily read with the free docs online to learn to use emacs.
<usney>I got it in the mail today.
<rekado>usney: you can view permissions on individual files with “ls -l the-file”; I don’t know of a tool that will check all of your files’ permissions and flag those that are “weird”.
<rekado>usney: neat!
<usney>plus the ereader allows me to read html books online too.
<rekado>usney: GNU programs usually come with manuals in info format that you can read directly in Emacs.
<rekado>with the index feature you can find relevant sections in the manuals much more quickly than with a web search even.
<rekado>an ereader is nice, though. I’ve got one too, and I did use it in the past for reading documentation that was not available in Info format.
<usney>yes I know but it is nice to use a separate text book that isn't on the same screen.
<usney>it is a text book not a manual
<usney>I use the manuals just fine.
<rekado>usney: I see. If you plan on learning Emacs I can recommend this book: https://www.masteringemacs.org/
<rekado>I think it is very well written and leads the way to very helpful features that might be obscured by strange defaults.
<usney>thank you
<Blackbeard>usney: emacs is hard but fun :)
<Blackbeard>At least for me it was
<leoprikler>It took me a weekend to get used to the keybindings. Now it takes effort not to expect them elsewhere.
<rekado>It took me *three* attempts to use Emacs productively. I only stuck around with it the third time because I was also switching from Qwerty to Dvorak, so my vim muscle memory was obsolete and I thus had no easy way to go back to vim.
<rekado>I always found Emacs weird. Now I do almost everything with Emacs.
<leoprikler>Emacs *is* weird. It's just that after some time you embrace your own weirdness and start using it.
<rekado>it took me some time to accept that I don’t *have* to use the defaults for everything. The more I moulded it after my habits the easier it became to use it.
<amartens[m]>Yeah it took me like around a year to get faster in Emacs than in any other editor ... now after 3 years of using it and around a 1000 lines of config, I can't go back ^^ I wouldn't even accept a job which wouldn't let me use Emacs.
<amartens[m]>rekado exactly, just like woth Lisp where you model the language on the problem, one can customize Emacs to oneself.
<amartens[m]>*with
<leoprikler>I once had one, where the software you're using was very closely monitored, but they still relied on the GNU tools very much.
<leoprikler>Which is why I got the (then still pretty recent) Emacs version 25 to run there and just snuck most of the Elisp stuff I needed past any security ;)
<civodul>cbaines: you once sent a patch that would fiddle with (@@ (ice-9 popen) open-process)
<civodul>that private procedure could disappear real soon
<amartens[m]>leoprikler: haha, awesome ^^
<civodul>cbaines: ↑ https://lists.gnu.org/archive/html/guile-devel/2020-03/msg00016.html
<civodul>what would be the impact for you?
<civodul>found it: https://issues.guix.gnu.org/issue/34638
<civodul>actually the latest iteration doesn't use (ice-9 popen) internals anymore
<usney>how often should I use "guix pull" rekado
<usney>is guix pull the same as apt update?
<usney>or maybe apt update; apt upgrade
<cbaines>civodul, I'm not currently using that code, although I'd like to
<civodul>yes, damn it, i dropped the ball again
<cbaines>I tried using it for the https://guix-patches-data.cbaines.net/ instance of the Guix Data Service, but I ran in to an odd error, and haven't looked at it yet
<cbaines>civodul, Not having that merged hasn't been blocking me, I've been mostly focusing on the data.guix.gnu.org instance, which has less of a need for it
<civodul>yeah
<leoprikler>usney: Pull regularly, but not obsessively.
<leoprikler>Checking every 15 minutes like certain Debian-based distros will not be needed ;)
<usney>like once a day leoprikler
<leoprikler>sounds better
<usney>cool
<usney>which is the best way to automate that? Is it still cron?
<usney>write a shell script and use cron to automate it.
<usney>is cron still a good choice?
<guix-vits>leoprikler: "checking every 15 minutes" -- why?
<leoprikler>idk, some distros do that
<usney>yes it is the new rule like the you can still eat it 5 second rule
<usney>ultra secure 15 minute internals debian fork
<guix-vits>hm... remembering that in Debian one can just switch off all updates but security related ones. Rock-solid, all of that.
<usney>what is thunderbolt?
<guix-vits>wikipedia is fine, but sometimes it's really easier to ask. loooong summaries. details.
<usney>does npm or pip repos have any non-free software?
<rekado>usney: yes
<usney>is there a way I can look up the license via the command line before I install?
<usney>flatpak allows me to do this
<guix-vits>usney: probably there is a way. this package claims to show the licenses of the packages and their dependencies: https://www.npmjs.com/package/license-checker
<guix-vits>notice "License: BSD-3-Clause" on the page.
<leoprikler>usney: `guix show <package`, then click on the license link
<leoprikler>s/<package/PACKAGE//
<guix-vits>+ check out recsel from recutils package (as manual suggest).
<usney>how do I install a webapp in icecat?
<guix-vits>i've the term-auto service stopped. how can i know why?
<Veera>Hi guix
***Veera is now known as Veera_
***Veera_ is now known as Veera
<guix-vits>Hi Veera_ Veera VeErA
***Veera is now known as Veera_
<efraim>sneek later tell janneke I'm building powerpc-linux-gnu bootstrap-tarballs on core-updates, we'll see how they go
<sneek>Welcome back efraim, you have 1 message!
<sneek>efraim, lfam says: How do you decide which Qt modules go in pyqt, efraim? Is there a list of required modules? Can we add more? I'm packaging something that expects to be able to run qtcharts from pyqt
<sneek>Okay.
<efraim>sneek: later tell lfam at the time it was whatever it seemed to want. we kept the qtwebkit one separate to keep the size/rebuild time down. feel free to add qtcharts or whatever, I think it spits out a list at build time
<sneek>Got it.
<efraim>sneek: botsnack
<sneek>:)
<efraim>oops, have to fix libffi on powerpc
<mbakke>janneke: the results are in!
<efraim>i'll see about fixing libffi for powerpc, if jonsger/ppc64 needs it (and probably does) we can grab all 4 of debian's patches and smush them together
<pkill9>my puloseaudio closed again, and i have a log of output this time, anyone can see why it shutdown? https://paste.centos.org/view/raw/003add56
<rekado>pkill9: is your audio device going to sleep?
<rekado>this looks like the beginning of the end: D: [pulseaudio] sink.c: alsa_output.pci-0000_00_1b.0.analog-stereo: state: IDLE -> UNLINKED
<janneke>mbakke: \o/
<sneek>janneke, you have 1 message!
<sneek>janneke, efraim says: I'm building powerpc-linux-gnu bootstrap-tarballs on core-updates, we'll see how they go
<janneke>mbakke: do they match what i pushed to wip-hurd?
<janneke>efraim: \o/ -- making great progress here
<efraim>janneke: looks like ppc(64) needs the last 4 commits here for libffi https://github.com/libffi/libffi/commits/master/src/powerpc
<pkill9>rekado: I'm not sure, how do I find out?
<janneke>efraim: okay...can we just apply them? /me is unsure if he should know something more about this...
<mbakke>janneke: guix hash -rx /gnu/store/lhca65c997pvic5cfrpm0dasniwqlg2a-bootstrap-tarballs-0 returns ...
<mbakke>drumroll 07jnq2by98f2a45k8wd2gj62iazvwfa4z7p3w3id4m1g0fdsvc3b
<efraim>janneke: I was going to merge them together and apply them all on top of core-updates, but it doesn't look like you'd need them
<efraim>so my suggestion is to continue on and i'll worry about ppc
<janneke>mbakke: \o/ -- me too
<janneke>ah, that could/should be added to the commit message
<mbakke>janneke: perfect :) I also unpacked the tarballs and did some sanity checks on the executables, seems fine
<mbakke>janneke: yes please :-)
<mbakke>I think we are good to merge then \o/
<janneke>efraim: good thanks -- and then we make way for efraim and ppc
*janneke rewrites commit message and remembers why his laptop is so hot and noisy
<janneke>mbakke: i was double checking a native build on the Hurd of `hello'
<janneke>i am currently building glibc-final; python-boot0 and gcc-7.5.0 are already built -- so i'm pretty sure we're fine
<guix-vits>Are someone know what 'v' stands for in "ter-v24b.psf.gz"?
<janneke>in an hour or so we know the anwser :-)
<rekado>efraim: why should we not build the GTK3 backend for youtube-viewer?
<efraim>I should copy the patch over to my powerpc laptop and fix libffi there too on pkgsrc
<rekado>I mean “frontend”
<efraim>rekado: it needs perl-gtk3 which isn't packaged yet. without it launching gtk3-youtube-viewer just errors on the console
<efraim>I have some WIP for switching to gtk3 but there's some errors still
*janneke lingers a bit...installing glibc-final now
<pkill9>rekado: /sys/module/snd_hda_intel/parameters/power_save is '0', so i guess it's not idling?
<efraim>civodul: re gnupg and default pinentry, that's a lot of code for a 1 line patch, and couldn't we have changed 'DIRSEP_S "pinentry" EXEEXT_S' to 'DIRSEP_S ".." DIRSEP_S "pinentry" EXEEXT_S' instead?
<efraim>then it'd also use whichever pinentry is in the profile of that gnupg
<efraim>sneek: later tell lfam it looks like right now python-pyqt doesn't build the plugins for Qt Designer or qmlscene
<sneek>Okay.
<efraim>sneek: botsnack
<sneek>:)
<janneke>mbakke: pushed!
<Veera>in pkg using git-fetch what does the sha256 is for
<andydarcyjewell>hey guix! hope everyone is coping well with lock-down. Still on this Factor packaging... how do I refer to the source directory with copy-file, please?
<Veera>I mean how come to precalculate it
<andydarcyjewell>Will it be something like (assoc-ref source ...) or similar? I can't work out how to search for this :-(
<jayspeer>hello, I have a small problem with emacs from guix
<brendyyn>andydarcyjewell: i think something like (assoc-ref inputs "source")
<jayspeer>when using markdown-mode emacs crashes because it tries writing to store (no wonder it gets killed)
<jayspeer>I'm using guix on ubuntu (work laptop)
<jayspeer>I've verified markdown-mode works on emacs shipped with ubuntu
<guix-vits>Veera: you need `guix hash` or `guix download` (see the manual, guix hash -rx for git, svn and other VCS stuff).
<andydarcyjewell>brendyyn: does the (source ...) stanza get propagated to inputs automatically then?
<guix-vits>jayspeer: can you please specify, why do you think it's attempt to write to The Store?
<brendyyn>yes i guess so
<andydarcyjewell>I tried (assoc-ref source origin)) already but that didn't work, so going to try your suggestion now
<brendyyn>you will need #:key inputs
<jayspeer>quix-vits: I run emacs with strace and I saw an attempt to write something in the store path when opening a README.md file
<jayspeer>way to reproduce (I guess?) install emacs and emacs-markdown-mode from guix; run emacs -q; open markdown file; let it crash :)
<efraim>janneke: congrats!
<janneke>efraim: thank you!
<efraim>do you know if it's possible to use --system=i586-gnu to build stuff on linux?
<efraim>also don't forget to update the manual :)
<andydarcyjewell>is there a way to output messages to the console from a package? just tried (display ...) but it didn't work.
<efraim>there's (format #t "text here"), or there's always (invoke "echo" "text here")
<andydarcyjewell>efraim: thanks
<mbakke>efraim: --system=i586-gnu would only work if Linux learns how to run native Hurd code
<jayspeer>sorry, something kicked me out. has anyone managed to test my problem?
<efraim>mbakke: I got a segfault error 11, so it doesn't look possible. maybe with the qemu-binfmt-service-type service
<amartens[m]>jayspeer: works for me on guix system
<guix-vits>amartens[m], jayspeer: WFM -- classic :)
<guix-vits>amartens[m]: now try to install Ubuntu... and emacs... :P
<andydarcyjewell>(string-append "source: " (assoc-ref inputs "source") gives me the downloaded tarball, how do I reference the directory into which it was unpacked, please?
<andydarcyjewell>oops, sorry, i included a bit of extraneous debug junk there
<andydarcyjewell>i meant just (assoc-ref inputs "source")
<amartens[m]>guix-vits: well he asked if someone has tested it :D I did, but on Guix System and was not able to reproduce his error. Maybe I can try in a vm with Ubuntu later today ...
<mbakke>andydarcyjewell: most build systems will step into the directory to which the source was unpacked
<mbakke>i.e. $PWD is the directory you are looking for
<andydarcyjewell>mbakke: ok, so is that a guile object then, or can I use that as the destination for copy-file ?
<efraim>'cwd' is something you can use for 'current working directory'
<guix-vits>amartens[m]: ++respect
<mbakke>andydarcyjewell: (copy-file (assoc-ref inputs "foo") ".") will copy the foo input to the current working directory
<jayspeer>I can imagine this will break on any foreign distro, but for it specifically breaks on ubuntu :/
<jayspeer>me*
<mbakke>jayspeer: do you know what file it tries to write?
<jayspeer>I'm trying to decipher it from strace, but I'm really fluent at strace messages. I can post trace if you want?
<mbakke>jayspeer: sure
<andydarcyjewell>mbakke: I think being in such an alien (for me) context as guile makes me forget what I already know :-(
<mbakke>andydarcyjewell: hehe, you'll get used to it ;-)
<andydarcyjewell>I've kind of reverted back to "novice" mode, despite being well accustomed to linux, bash, and quite a few other programming languages
<jayspeer> https://paste.debian.net/1136762/
<jayspeer>this is tail of the trace if you need more, pls ask
<mbakke>jayspeer: weird, it looks like it segfaults right after trying to open /home/jayspeer/.guix-profile/share/fonts/opentype/Inconsolata.otf which does not exist
<mbakke>perhaps your fontconfig cache is stale?
<mbakke>try running 'fc-cache -rv' in that case
<jayspeer>huh, that would extremely sad reason for a segfault
<jayspeer>you know what? It works :')
<mbakke>O_O
<jayspeer>I assumed this was just an artifact since I have Inconsolata installed
<jayspeer>thank you very much! :)
<mbakke>I wonder what it would take to coordinate an effort between all the distributions to fix the font situation for good..
<mbakke>for starters a search path would be nice instead of needing the fontconfig cache
<mbakke>jayspeer: glad to help! :)
<civodul>efraim: DIRSEP_S and EXEEXT_S don't make sense on GNU/{Linux,Hurd}, that's why i removed them
<civodul>using the pinentry from the same profile would be nice, but how would you do that?
<civodul>i guess we'd need a new function to determine what the current profile is
<civodul>similar to current-profile in (guix describe)
<civodul>but maybe that's a bit overkill?
<efraim>looking just at the snippet in the patch it looked like it was looking in the current directory for pinentry
<efraim>so my suggestion wouldn't have worked anyway
<Veera>Does guix download saves also the uri in store
<efraim>but I'm guessing it was looking in the same directory in the store, not in the profile
<civodul>ah yes
<rekado>Veera: yes.
<Veera>some days ago someone suugested that if transfers get interrupted often due to networks issues then download using wget and do guix download localfile and use it
<Veera>Then is it possible to specify the local file in pkg origin spec
***amartens[m] is now known as starcrATI[m]
<mbakke>Veera: guix will check if the store contains a file with the expected name and hash before attempting to download it
<mbakke>so you don't have to do anything special with the origin
<Veera>ok that explains things
<guix-vits>mbakke: what is the best way to get hash for (git-fetch (it's also Veera 's original question)?
<Veera>guix hash info gave a example
<Veera>it said checkout git clone git://url
<Veera>and use guix -rx that dir to get hash
<Veera>will it turn out same each time if that dir is not touched mean left to same commit
<Veera>or will it return a bunch of hashes
<raghavgururajan>rekado I don't the linphone-desktop has cmake script that has a line to use pkg-config for finding pc files. Despite the fact that bctoolbox's source tarball has bctoolbox.pc.in file.
<anadon>Good morning Guix!
<guix-vits>Veera: i just hoped that there is a way to do `git clone` once (to the results ended up inside of The Store). Probably unchanged dir will return same hash (idk :).
<Veera>Am pkging a GUI app which provides a separate library and needs it. Where should I put those?
<rekado>raghavgururajan: does the bctoolbox.pc get *installed*?
<rekado>raghavgururajan: there is no need for an explicit call to pkg-config in CMakeLists.txt
<rekado>Veera: depends on the library.
<rekado>Veera: for now I wouldn’t worry about it and put it above the definition for the GUI package.
<rekado>let the reviewer sort it out :)
<Veera>were do GUI calculators go in which scm or should they be separate?
<raghavgururajan>rekado Yes, there is "/gnu/store/m92m6bg0xcl28djxg8h97sszf3gdl42r-bctoolbox-4.3.1/lib/pkgconfig/bctoolbox.pc".
<raghavgururajan>rekado So it is installed.
<rekado>excellent. What does it contain?
*rekado –> groceries
<guix-vits>Veera: `guix search calculator|less` shows that mate-calculator go to mate.scm, gnome-calculator to gnome.scm ... Probably "it's depends".
<raghavgururajan>rekado https://bin.disroot.org/?838aecf46c440cc3#83ysBj6AHCLhtypzfxoD9vH9VGVcdUHfG7FcfBywZ6kZ
<rekado>Libs.private is truncated
<Veera>guix-vits: this calc app is an independent one not tied to DM
<raghavgururajan>rekado Oops! Here, https://bin.disroot.org/?c521162231941709#FRoTNFdFo3iyYZgr94MuAu1pjz2FcZzRjFDC7rYpmLRY
<chiefgoat>vagrantc: I've actually been holding off on Guix until the PBP was supported as that's my daily driver laptop. Anything special I need to use your kernel?
<guix-vits>Veera: some calcs go to math.scm, some to algebra.scm ... some in admin.scm (IP-calc)...
<anadon>Why does guix use a daemon rather than a program with an 's' bit set?
<Veera>guix-vits: quite a lead
***Server sets mode: +cnt
***matijja``` is now known as matijja`
***matijja` is now known as matijja
<civodul>anadon: partly for "historical reasons", but also because it allows for better sharing & scheduling
<raghavgururajan>linphone-desktop now wants qt5. Oh boy!
<mbakke>civodul: nice find wrt hard links in glibc-locales
<mbakke>as the package is really quite small, WDYT of simply wrapping guix-daemon and the pulled guix with GUIX_LOCPATH so users never have to care about it again?
<civodul>mbakke: that's already done :-)
<civodul>but they embed glibc-utf8-locales, not glibc-locales
<anadon>civodul: Do you or other devs see merit in a daemonless, event driven only model for guix? I know for my work, having that and avoiding having a service running would be a positive.
<anadon>work -> workplace/job/pays me to do things
<civodul>anadon: there's interest for this on HPC clusters
<civodul>is this also the context you're operating in?
<anadon>Incidentally, I'm thinking of our HPC cluster.
<civodul>yeah
<civodul>i think the whole "setuid-root binary is fine but daemon is not" strategy is questionable
<civodul>but it's widespread so we can't ignore it
<mbakke>civodul: wait, what? people ask about GUIX_LOCPATH issues every day on this channel and I can't find the commit
<anadon>For me, it is a design pattern that is intuitive. "Event Driven" design everything.
<civodul>though you cannot achieve a setup like https://hpc.guix.info/blog/2017/11/installing-guix-on-a-cluster/ without a daemon
<civodul>mbakke: the 'guix' package does wrap-program, and (guix self) does something similar
<mbakke>huh :)
<mbakke>so I guess the users that still have problem are just running an old daemon
<anadon>I've read it, but I'm one of the more technically adept in our group and I still have my butt handed to me with guix. I'm not prepared to just up and demand that, and the HPC cluster technically isn't my job.
<civodul>mbakke: could be, we'd need to find out, or it's going to be hard to improve on what we have :-)
<civodul>anadon: sure, so what kind of setup are you aiming for?
<anadon>It was just something I was thinking about this morning and it struck me that I just had no clue why.
<civodul>ok
<anadon>There is still a lot I don't understand about guix, so every now and again I'll just have a question.
<civodul>sure, don't hesitate to ask!
<anadon>How does the daemon enable better scaling and scheduling than a 's' bit program which has IPC? What are the historical reasons?
<civodul>there can several clients who want to build a given set of packages concurrently
<civodul>that needs coordination
<civodul>if you spawn N independent programs, they have to find out which resources are available etc.
<civodul>conversely, the daemon has a global view
<anadon>What about coordination through shared memory?
*mbakke did `guix pull --branch=core-updates` and got locale warning with LANG set to en_US.UTF-8
<anadon>Or coordination through opening a unix port, then checking binding status and acting off of that?
<civodul>mbakke: can you paste the exact command and output?
<civodul>anadon: what do you mean by "Unix port"?
<civodul>there are certainly other ways to approximate that behavior
<civodul>but it's easier to have a centralized orchestrator
<anadon>civodul: Mixed terms a bit. I think this: https://en.wikipedia.org/wiki/Unix_domain_socket
<anadon>There could be some merit in having the daemon exist on demand, and then not require integration with hurd or systemd.
<mbakke>civodul: I did `guix pull --commit=dd0568a77951804d97884a0a0a3a3ed38ca0a184 --branch=core-updates` (a week old, perhaps there are recent changes from 'master'?)
<mbakke>then `guix describe`, which prints: 'guile: warning: failed to install locale'
<mbakke>and then the expected output, of course
<civodul>mbakke: what if you do: "guix repl" and (setlocale LC_ALL "")?
<Veera>hey
<civodul>mbakke: the warning emitted by Guile cannot be avoided because the wrapper in (guix self) is itself Guile code
<civodul>in 3.0 i proposed to remove that warning in Guile itself
<Veera>After sending a patch for adding a pkg do I have to send a message to guix-devel list?
<civodul>hi Veera
<civodul>Veera: no, someone will pick it up from guix-patches, time permitting
<civodul>that person will email you if they have comments or if they've applied the patch
<Veera>hi civodul
<Veera>ok
<Veera>april 7 is the deadline for outreachy
<andydarcyjewell>Sorry if I'm repeating myself here, but it seemed like the web chat page had stalled, so re-submitting my question now.
<andydarcyjewell>Surely I must be nearly there now! How do I pass arguments to programs with invoke? Silly example: If I do (invoke "ls") I get a directory listing, but if I do (invoke "ls somefile") I get error 127. I don't really want to list files - I want to run a program with some parameters.
<Veera>yes dannym commented and I sent v2 patch
<Veera>but xchat patch i sent no comments yet
<civodul>Veera: so hopefully Danny will reply shortly, or otherwise someone else will follow up
<Veera>Gabor Boskovits told me make it as wontfix
<civodul>i understand there's a deadline, we'll try to be quick
<Veera>that will be nice
<anadon>civodul: I am all for that patch to remove the LANG warning.
<roptat>andydarcyjewell, with more arguments: (invoke "ls" "somefile")
<Veera>Lafreniere Joseph sent a reply mail for XChat which I posted to guix-devel
<andydarcyjewell>roptat: I thought I'd tried that already, but will give it a go
<roptat>of course if you need more than one argument, you need even more strings: (invoke "ls" "-l" "-a" "somefile")
<roptat>(which is silly: (invoke "ls" "-la" "somefile") workes as well :))
<guix-vits>andydarcyjewell: `cd guix/gnu/packages; grep -l '(something-smth' *;done`
<Veera>civodul: current iso img has the bug that it randomly comes to first page or the second after next many steps or sometimes it skips the steps in series and jump x forward
<guix-vits>Veera: in "Graphical Install"?
<Veera>guix-vits: yes in graphical install
<Veera>As I was new I went with graphical install
<Veera>When it did not worked I switched to VM
<Veera>Then I realized I was stuck because of lack knowledge
<andydarcyjewell>roptat: I've obviously lost track of the things I've tried. Obvious really, and works.
<Veera>So again I tried with iso but this time manual one; I got success
<guix-vits>Veera: yes, it apeared broken for me too. But the .iso is nearly one year old. For it's age it's works surprisingly well hehe.
<Veera>The moment the site describes Guix as different concept OS and I thought it be entirely different from other Linux distros.
<Veera>But it is simple tui interface
<Veera>but it uses Linux kernel and all it's features drivers, fs, net etc
<Veera>Only Guix pkg manager/daemon/scheme is different
<Veera>and uses gnu/unix/linux softwares.
<Veera>Only if the docs were explicit it would have been very easy.
<rekado>Veera: I don’t understand what you are suggesting. Guix is a very different kind of operating system, but it is a GNU/Linux system.
<Veera>rekado: Well GNU/Linux means uses GNU software and Linux means uses Linux kernel
<Veera>and thats make it same with other distros
<Veera>and already said which makes it a different one
<vagrantc>guix is more different than most other gnu/linux distros
<guix-vits>Veera: "if the docs" ...and if the error-reports from guile were better: i'm almost gived up trying to adjust the 'console-font-service-type' in %base-services with (modify-services.
<vagrantc>but yes, the fundamental difference is the package manager
<rekado>guix-vits: please report problems with the docs to bug-guix@gnu.org
<vagrantc>though there's work to run guix on hurd instead of linux
<Veera>I mean the installation is same(abstract one); configuring devices and setting users, making partitions ...
<rekado>(“GNU/Linux” means that this is the GNU system using Linux as the kernel; it does not mean “uses GNU software”)
<Veera>So there is difference between GNU system and uses GNU software.
<Veera>I did not know
<vagrantc>builder for `/gnu/store/zrbdf0a4xs9h9f63x4q4qzpj5h3kp1z8-icu4c-64.2.drv' failed with exit code 1 ... poor armhf
<Veera>RMS used to say that since Linux kernel uses GNU infrastructure(software for bootstrap...) and user programs so it should be called GNU/Linux and not Linux
<Veera>sorry RMS advocates and not used to say
<guix-vits>Veera: OS is GNU Linux, Kernel is Linux.
<guix-vits>and Guix uses Linux-Libre, due to many blobs in vanilla Linux.
<Veera>that I know
<Veera>Is it that Guix follows GNU/RMS and may be FSF
<rekado>Veera: the argument is about the name of the system, which is often called “Linux”
<vagrantc>guix is an official gnu project
<rekado>Guix does not follow RMS.
<rekado>Guix is part of the GNU project.
*rekado –> supertuxkart
<Veera>Ok Guix is one of GNU project lead by RMS and others
<Veera>GNU is lead by RMS and others
<rekado>founded by RMS, yes.
<Veera>Founded and I think still chief
<Veera>He has not left it as he left FSF because of controversy
<vagrantc>it's perhaps not the most important thing to get caught up on
<Veera>okay
<starcrATI[m]>how can I switch my system to use the Hurd once it's available ? :) will I need to reinstall ?
<vagrantc>guix is a community of people helping people :)
<vagrantc>starcrATI[m]: probably? the binaries are all basically incompatible
<Veera>ah, same or similar things said by libres
<Veera>As I know Hurd status is poor
<Veera>Debian had, may be has Hurd support
<vagrantc>there's active work on porting the hurd in guix
<Veera>that's good
<vagrantc>debian's port is still very much a work-in-progress ...
<civodul>although it actually works!
<civodul>as in you can get Xorg running, and GNOME, and whatnot
<starcrATI[m]>vagrantc okay, thanks :)
<starcrATI[m]>civodul can I already use it today ? :)
*vagrantc wonders how plausible it would be to dual-boot guix linux-libre and guix hurd from the same machine
<vagrantc>just have different generations happening to be different operating systems? :)
<civodul>starcrATI[m]: yes, it's just lacking features (SMP, 64-bit, drivers), but otherwise it's undistinguishable from GNU/Linux at first sight
<civodul>vagrantc: yeah i wonder too!
<starcrATI[m]>vagrantc: xDD why not
<vagrantc>seems plausible...
<vagrantc>could maybe even do that with x86_64/i686 now, no?
<Veera>what's xDD
<starcrATI[m]>> civodul: starcrATI: yes, it's just lacking features (SMP, 64-bit, drivers), but otherwise it's undistinguishable from GNU/Linux at first sight
<starcrATI[m]>ah okay, tnanks
<starcrATI[m]>*thanks
<civodul>vagrantc: we'd need some magic so that the right profile is selected (and built?) when we boot
<vagrantc>doesn't grub already provide that?
<vagrantc>oh, you mean for end-users?
<vagrantc>user profiles
<Veera>some distro's are providing x86_64/i686 function
<Veera>multilib thing
<Veera>thanks. bye
<starcrATI[m]>vagrantc; how is performance on hurd?
<vagrantc>starcrATI[m]: haven't tried in decades :)
*vagrantc just watches from the sidelines
<starcrATI[m]>okay :D
<guix-vits>decades..
<vagrantc>almost... :)
<vagrantc>~18 years, maybe?
<vagrantc>sounds like it's about time to try again :)
<user_oreloznog>I installed hurd K9 in the beginning of 2000 years
<user_oreloznog>Maybe 2004
<starcrATI[m]>* feels young
<guix-vits>interesting, how old rekado? -- others speak with him, but for me it's just some *other* English.
<starcrATI[m]>* * starcrATI feels young
<guix-vits>then when systemd will be ported to Guix, there weel be Oldix from "veteran Guix admins".
<guix-vits>eww.. scarry.
<guix-vits>:P
<drakonis>haw
<drakonis>i'd like to point out that the reason devuan is called is due to veteran unix admins
<drakonis>vua + debian = devuan
*nckx ♥ Devuan.
<nckx>It's a great honeypot to keep the ‘veteran’ ‘unix’ ‘admins’ away from everyone else.
<nckx>Good morning, Guix.
<drakonis>quite
<guix-vits>Hi nckx.
<nckx>Hi!
*jonsger runs a distribution of devuan on is phone :P
<nckx>guix-vits: You sent me a dot. It was a very nice dot (.) but I do not know what to do with it.
<guix-vits>i wanted to know if you're alive. Last commit was near three days ago.
<nckx>😃
<nckx>I've been quite ill.
<guix-vits>drakonis: in russian this sounds (or may sound) like divan (soft sitting, probably "coach")
<guix-vits>if this is deev one, then no. But jokes about that it's one of the best names (and simultaneously a motto) were told on some forums.
<aitzkora>Hi! I want to explore the tree of compilation when a guix failed, but the message give me only a path to a big drv file ?
<guix-vits>Hi aitzkora, this it a mistery for me. Try ask on #guix (currently you in personal messaging with only me).
<guix-vits>sorry, i'm glitching.
<guix-vits>i just switched screen and seen a date on the top...
<aitzkora>guix-vits: no problem!
<NieDzejkob>It should also show you the path to a log file
<nckx>aitzkora: Not sure what you mean by ‘tree’, but ‘guix build PACKAGE…’, at least, will print the path to a build log as one of the last lines.
<NieDzejkob>(.log.bz2, see it with bzless of bzcat)
<bricewge>Hey Guix!
<guix-vits>aitzkora: you want to preserve the failed build dir?
<nckx>Ah, that makes sense.
<aitzkora>nckx: but it is a temporary path destroy after compilation isn'it
<nckx>-K will keep it and print it.
<aitzkora>ncks ! thanx
<bricewge>How can I help https://issues.guix.info/issue/35305 (add lightdm service) making it master?
<bricewge>Does it make sense reviewing it even tough I don't have right to the git repo?
<nckx>bricewge: Reviewing always helps, even if you can't give the final LGTM.
<nckx>Especially if you're familiar with lightdm (I'm not).
<civodul>aitzkora: see https://guix.gnu.org/manual/en/html_node/Debugging-Build-Failures.html :-)
<aitzkora>civodul: it worked with -K !
<civodul>yay, awesome :-)
<bricewge>nckx: Will do. Familiar maybe not, I was just using it NixOS before.
<aitzkora>civodul: tests phase do not fail when I compile on my arch, but appears with guix :-)
<kawerte>Hi everyone
<kawerte>Can someone help me? I want to import packages from nix, but am unable to do so.
<drakonis>you need the daemon to be running
<drakonis>though the importer might function only for basic packages
<kawerte>so if I want to install say firefox it won't work?
<guix-vits>kawerte: Guix has IceCat, which is refined FireFox.
<kawerte>it has problems
<drakonis>i don't think they'll help you out with running nix for dodging the lack of software considered non free
<kawerte>and I am web developer, gonna need chrome and firefox
<drakonis>icecat has strange media issues
<drakonis>not to mention that chromium ungoogled also has issues
<kawerte>sadly there is no distro that is based on lisp and is not gnu...
<kawerte>so I wanted to make the best of what there is available
<guix-vits>kawerte: i'm mean, there is some "frame", see? You can start with IceCat definition (gnu/packages/gnuzilla) and roast you own fox :)
<drakonis>kawerte: you can set up nix by the way
<kawerte>yeah
<kawerte>already did that
<drakonis>you have to set up the channels you want
<kawerte>yeah, already set it to the unstable branch
<drakonis>cool
<guix-vits>kawerte: what about qutebrowser?
<kawerte>not sure how rolling back is going to work though
<kawerte>isn't qutebrowser vim keybindings based?
<guix-vits>yes, but it has Chromium engine (inspector works)
<drakonis>not even the current one
<kawerte>maybe I will have to roll-back two times when needed? once for guix and once for nix?
<drakonis>nix doesnt drive your system here, guix does
<kawerte>if I roll-back on guix then, what happens with my latest installed packages from nix?
<kawerte>do they get removed as well?
<drakonis>doesnt change
<drakonis>if you do rollback guix to a generation before you installed nix, you only lose the daemon
<kawerte>hmm, that makes sense. sorry for the dumb question :)
<kawerte>that should be the solution then I guess, whatever isn't available in guix I can install from nix
<kawerte>thanks @drakonis @guix-vits
<drakonis>its not exactly the right way to do things here
<drakonis>and there's things that might not work very well with nix only
<joshuaBPMan>Hey guix! I've just finished librebooting my T400, and I'm having weird ethernet issues.
<sneek>Welcome back joshuaBPMan, you have 1 message!
<sneek>joshuaBPMan, nckx says: I've disabled the failing test on master and've pinged upstream about https://paste.debian.net/plain/1135869
<nckx>Oh, that's old.
<joshuaBPMan>I suppose so. ifconfig -a shows my ethernet device
<joshuaBPMan>enp0s29
<joshuaBPMan>dhclient -v enp0s29 fails to create a connection.
<joshuaBPMan>"unable to set up timer: argument out of range"
<aitzkora>I do not remember the command to compute the checksum (without using download)
<bavier`>aitzkora: 'guix hash'
<aitzkora>bavier`: yes! with -rx :-)
<aitzkora>thx
<lfam>Howdy
<sneek>Welcome back lfam, you have 2 messages!
<sneek>lfam, efraim says: at the time it was whatever it seemed to want. we kept the qtwebkit one separate to keep the size/rebuild time down. feel free to add qtcharts or whatever, I think it spits out a list at build time
<sneek>lfam, efraim says: it looks like right now python-pyqt doesn't build the plugins for Qt Designer or qmlscene
<raghavgururajan>Folks!
<raghavgururajan>I get some Qt related errors, during the build of "linphone-desktop".
<raghavgururajan>LOG: https://bin.disroot.org/?79c1cc2131f7235f#HHS3xWeLCKd98L1fqZB6xEwnkPQWqyJXpoemhwppzRNV
<raghavgururajan>DIFF: https://bin.disroot.org/?5c92968202a11fe5#5TrosGgDg5i5ZPve4VXr452wTrNHXaiHo146uc7gzZXP
<raghavgururajan>I am not able to understand what those error mean. They kinda have flags in it.
<guix-vits>s/folks/brave hearths/
<jonsger>janneke: is core-updates now build on CI?
<janneke>jonsger: i was just asking myself that question too
<mbakke>jonsger, janneke: still only building a small subset, now waiting for Guile 3.0.2! and *then* we'll start the full rebuild, promise! :P
*mbakke works on some last-minute updates, as usual...
<joshuaBPMan>mbakke: what is new in guile 3.0.2 ?
<mbakke>joshuaBPMan: quite a lot of bugfixes, some related to gc, and one fixing this ancient bug: https://bugs.gnu.org/28211
<mbakke>also some new SRFI's :)
<joshuaBPMan>mbakke: oh! That's pretty cool!
<NieDzejkob>How are c-u rebuilds on CI controlled? Do you make builds of core packages fail on purpose, or is there some less barbaric control?
<janneke>mbakke: right, that makes some sense -- still, with the glibc updates -- ah no, no significant updates since yesterday!
*janneke hopes you all don't mind me rubber-ducking you ;-)
<mbakke>NieDzejkob: I don't understand the question, the job specification is stored in a sqlite database
<mbakke>janneke: I enjoy being a rubber duck!
<lfam>I think they mean, how do you build only the core subset of packages?
<mbakke>lfam: ah thanks, I get it now. The specification has to be updated directly in the SQLite database.
<lfam>Is it separate from the cuirass-jobs.scm in maintenance.git?
<mbakke>lfam: they are interconnected, the SQLite specification evaluates that file and currently passes 'subset as argument to hydra-jobs, which builds only %core-packages
<lfam>Gotcha
<GilbertVolcano>Hey guys, I install Guix using the binary on top of a Ubuntu install. But when I reboot it still boots to Ubuntu. Can anyone help?
<joshuaBPMan>GilbertVolcano: When you install that way, you will still use Ubuntu.
<joshuaBPMan>You essentially installed guix as an additional package manager.
<joshuaBPMan>If you want the Guix Distribution, then you need to install Guix System.
<joshuaBPMan>Let me send you some links to help you distinguish between guix the package manager and guix System.
<GilbertVolcano>How do I do that?
<GilbertVolcano>I tried using the USB install but it didn't work.
<lfam>Now that you have the Guix package manager installed, you can use the `guix system init` command
<lfam>It would be good to know why the system installer did not work
<lfam>What happened?
***joshua is now known as Guest90983
<joshuaBPMan>I did not realize that you could use the guix system init command to turn a foreign distro into Guix System. That is kind of cool!
<GilbertVolcano>When I boot is says "Welcome to GRUB! error: Unknown file system. Entering rescue mode grub rescue>"
<lfam>Yes, `guix system init` can be used to take over a running system, or in an installer setting. It will not get rid of the existing system but you can't really expect the Ubuntu installation to work correctly afterwards
<lfam>GilbertVolcano: My guess is that your config.scm had some mistakes in the bootloader and filesystems sections
<lfam>If you share it with us on <https://paste.debian.net> we can try to help
<GilbertVolcano>I used rufus to make the USB boot
<GilbertVolcano>Okay
<lfam>Oh, did you mean that booting the installer USB itself fails?
<GilbertVolcano>yes
<lfam>I see
<lfam>That's a different story
<lfam>I recommend following these instructions: https://guix.gnu.org/manual/en/html_node/USB-Stick-and-DVD-Installation.html#Copying-to-a-USB-Stick
<allana>Hi guix. Can anyone point me to an example package definition where a custom make target is specified? E.g. when the build phase would call "make mything" instead of just "make"?
<joshuaBPMan>GilbertVolcano: what laptop are you installing guix to?
<lfam>allana: Search the package definitions for '#:make-flags'
<allana>lfam: thanks!
<joshuaBPMan>GilbertVolcano: you may have to complete the installation manually. I personally have had a hard time getting the graphical install to work reliably.
<joshuaBPMan>You'll need to learn how to switch to a virtual console.
<GilbertVolcano>@joshuaBPMan I'm install it on a desktop I built myself
<joshuaBPMan>GilbertVolcano: ahhh. Ok. I've only installed it on laptops. If you have any questions feel free to ask.
<GilbertVolcano>I noodled around with the BIOS and got past the old error. Now I'm seeing this error "Enter ',help' for help scheme@(guile-user)>_"
<GilbertVolcano>thank you lfam
<joshuaBPMan>But, if you get a working usb stick, and it boots just fine, (hopefully) you may see the screen flash blue a lot. That means guix did boot properly, but the graphical install is not working so well. At that point you can switch to virtual console 3 and try the manual installation.
<GilbertVolcano>okay it recommends using dd, I didn't use that when creating the USB boot stick, I'll remake it and use dd to format it.
<joshuaBPMan>GilbertVolcano: It sounds like you are in the early boot scheme...That's pretty low level. I'm not certain how to help you there.
<lfam>Yeah, seeing the Scheme emergency prompt means it failed to boot
<lfam>I would just re-flash the USB and try again
<bavier`>ugh, I though Folding@Home was free software
<guix-vits>if one should change (uri to last "following redirection"?
<Blackbeard>bavier`: it is proprietary :(
<Blackbeard>bavier`: but they "open source some parts"
<lfam>guix-vits: Sometimes
<bavier`>Blackbeard: yeah, just kept getting more depressed the further I read in their faq
<Blackbeard>I don't understand why a school doing research to help the world and asking for help would need to keep the software proprietary. Makes zero sense to me
<bavier`>unfortunate, really
<guix-vits>lfam: thanks, then i'll prefer shorter one.
<bavier`>seems a bunch of it related to keeping the contests "fair" and minimize cheating
<Blackbeard>BOINC has coronavirus research and it is in GUIX
<bavier`>Blackbeard: good point
<Blackbeard>Although I couldn't make it work from parabola. Will try with guix system today
<Blackbeard>bavier`: they make no sense. The contest could easily be server side or something. I went for BOINC let me tell you which one is doing coronavirus research
<GilbertVolcano>I get the gray "guix" option splash screen, but after that I get a black screeen
<Blackbeard>bavier`: rosseta@home
<GilbertVolcano>I'm using an ATI card, could that be the problem?
<bavier`>Blackbeard: thanks!
<Blackbeard>bavier`: this is the URL http://boinc.bakerlab.org/rosetta/ the client will ask for it
<Blackbeard>If you can make it work please let me know
<Blackbeard>I couldn't
<GilbertVolcano>I think my graphics card requires non free drivers, how would I load those before booting the Guix USB installer?
<lfam>GilbertVolcano: I feel like it should be possible to install Guix without the GPU... in any case, if that is the problem, you'll need to avoid the GPU since Guix doesn't include non-free drivers
<lfam>If it were me, I'd try switching to another TTY and installing on the command line
<mbakke>ugh, core-updates has a huge merge conflict with 01e30442007a5d2b0f8c443896f7d6dc850b140f, because python-sortedcontainers is now used by pytest and friends
<joshuaBPMan>GilbertVolcano: I agree with lfam. Guix's graphical install doesn't always reliably work. You may need to use the manual install.
<mbakke>I think I'll revert that commit as part of the merge and try to sort it out afterwards
<lfam>mbakke: You might need to revert the whole series
<lfam>How is it a huge merge conflict? It seems like a relatively self-contained change
<mbakke>lfam: it conflicts with 4c2530afce2f7f43795c217d43f3d66069896257
<mbakke>lfam: if I were to take the version from 'master' there would be a dependency cycle
<lfam>Hm...
<mbakke>because python-sortedcontainers is part of the pytest bootstrap
<lfam>I was about to suggest a bootstrap variant of sortedcontainers
<lfam>Python is full of things like this
<pkill9>i can't get boinc to work
<lfam>And for little benefit IMO, since the test suites are mostly quite fragile
<pkill9>it wont login to rosetta
<GilbertVolcano>"If it were me, I'd try switching to another TTY and installing on the command line" How do I do that?
<pkill9>and yea it sucks folding@home isn't free software, but there has been a surge in attention and so they may open source it
<mbakke>lfam: yes, we'll need a bootstrap variant... actually it's best to add it pre-merge
<lfam>GilbertVolcano: On most computers you can press CTRL+ALT+F[1,2,3,4,...] to switch to various TTYs
<nckx>pkill9: Really? Their excuse was always ‘ppl will cheat 4 points’.
<pkill9>oh ok, maybe not then
<bavier`>pkill9: I couldn't get rosetta@home to add either, just now
<nckx>Well no, I hope your source is better than my source and that's plausible, so I hope you're right.
<lfam>mbakke: I can work on it in a couple hours if you give me an idea of how the dependency graph should look
<pkill9>what are the 'points' for?
<GilbertVolcano>When I hit e at the Guix Installer black screen I'm taken to what I believe is the GRUB config scren.
<bavier`>pkill9: they're "internet points" afaik
<GilbertVolcano>In it is stays modprobe.blacklist=radeon
<pkill9>they should atleast provide a FOSS client without this points system
<GilbertVolcano>can I delete that and put in "nosetmode"?
<pkill9>arguably they should just provide a FOSS client
<Blackbeard>The only positivo thing
<Blackbeard>Is that I read they have had so much attention
<Blackbeard>They are running out of jobs to send
<mbakke>lfam: basically, each of the new dependencies would also need -bootstrap variants that does not use pytest.. I think it's easier to do it myself than explain the whole pytest bootstrap graph! :-)
<Blackbeard>And that the clients are not doing much now
<lfam>I feel like most of the people who won't run folding@home because it's not FOSS would still refuse to run a free client because it's a "non-free service" or some nonsense like that
<nckx>wat
<Blackbeard>But they absolutely should release the client with a free license
<Blackbeard>lfam: I would compromise for coronavirus sake
<lfam>Me too
<lfam>Okay mbakke
<Blackbeard>I did installed it on my mom's laptop with Ubuntu and some proprietary things
<Blackbeard>But in mine I didn't. I can compromise a bit.
<mbakke>there are also some really tricky merge conflicts in guix/store.scm, guix/scripts/environment.scm, guix/scripts/pack.scm, guix/scripts/package.scm, and guix/scripts/pull.scm. I'm aborting the merge.
<lfam>Sorry for being inflammatory, but I'm not surprised we haven't managed to convince folding@home to free the client
<lfam>Oof mbakke
<lfam>It's a shame that merge conflict resolution can't be distributed easily
<lfam>Unless I'm missing part of the workflow it's a place where Git really doesn't help much
<mbakke>lfam: good point, seems impossible short of screen sharing
<Blackbeard>bavier`: it didn't work then?
<lfam>mbakke: You'd have to make a "merge to-do list" and collect diffs from people working on various areas of the conflict
<lfam>Messy
<cbaines>When thinking about this before, I think a world where core-updates is just a collection of patch sets might work
<cbaines>you just create the branch from master + the patch sets that apply
<bavier`>Blackbeard: not immediately, no; I'm trying a few things
<cbaines>In that model you could automate continually updating core-updates from master, identifying and temporarily removing patch sets that don't apply
<bavier`>the log output contained: "BOINC can't access Internet - check network connection or proxy configuration."
<bavier`>which is weird
<mbakke>the last few merges have had about one conflict per 20 commits
<mbakke>cbaines: interesting, that is basically rebase on every merge, right?
<Blackbeard>bavier`: did you install the server package? I am not sure what it is for to be honest
<lfam>I've done that before cbaines, at least for my own patches that were intended for core-updates
<cbaines>mbakke, effectively, you'd "refresh" the branch as often as is useful, maybe once a day or once every two days
<lfam>Or very similar
<bavier`>Blackbeard: just the boinc-client package
<cbaines>but refreshing it could be automated, because you'd just exclude any sets of patches that don't apply cleanly
<GilbertVolcano>A "No UMS Support in Radeon Module" flashed by before the screen went black during my Guix install.
<GilbertVolcano>How do I manually install?
<mbakke>one way to distribute the merge conflict burden is to merge more often
<mbakke>I can avoid the conflicts in guix/scripts/* by merging the commit just before 041b340da409078951267b6a8c43b27716e6b7ec
<cbaines>mbakke, I'm all for merging core-updates more often. I want to get the Ruby 2.6 update that's on there!
<mbakke>I'd still have to solve the python-sortedcontainers issue, but could ask civodul to merge the later commits
<mbakke>actually, I think I'll do just that this round
<mbakke>cbaines: for the next core-updates, let's set a hard "freeze" date and stick with it
<mbakke>I don't think every 2.5 months will cut it any more because of the huge amount of new packages since that was written
<mbakke>2-3 times per year seems more achievable
<cbaines>well, thanks for all your hard work on it
<mbakke>given that glibc releases the first of every february and august I've considered to sync up with that, i.e. freeze and start merging the branch every march and september
<bavier`>Blackbeard: my only lead so far is that maybe boinc doesn't have the ca-certs it needs to authenticate with projects
<Blackbeard>bavier`: that may be it 🤔
<GilbertVolcano>How do I being a manual install of Guix. Do I open a command prompt in the Ubuntu OS i have already installed on the disk?
<mbakke>GilbertVolcano: you can start a shell from the graphical installer, do the partitioning on the command line, and then run 'herd start cow-store && guix system init my-config.scm'
<GilbertVolcano>How do I start a shell from the graphical installer. I tried hitting Ctrl+Alt+F6 but I just get the Guix Splash page instead of a prompt.
<Blackbeard>GilbertVolcano: the graphical installer will ask you if you want it
<Blackbeard>GilbertVolcano: just select language and a keyboard distribution first
<GilbertVolcano>I can't reach the blue graphical installer, it always cuts to black before then. If I had an NVIDA card would this work better?
<mbakke>GilbertVolcano: the current installer has problems with AMD graphics indeed
<GilbertVolcano>Oh darn, okay, so I should go get an NVIDA card?
<mbakke>GilbertVolcano: if you happen to have one lying around, that should work indeed
<lispmacs[work]>hey, is that Gnome 3.34 coming around the corner
<lispmacs[work]>excited about that, hoping it will fix my Evolution email problem
<GilbertVolcano>No, I'll need to go buy one, does it matter which NVIDA card I get.
<mbakke>lispmacs[work]: you should be able to pull a branch that has GNOME 3.34 in a few days
<NieDzejkob>GilbertVolcano: try Ctrl+Alt+F3 and F4 instead of F6
<NieDzejkob>and F2 for the docs IIRC
<GilbertVolcano>Oh that worked NieDzejkob!
<mbakke>phew!
<mbakke>lfam: good news! none of the sortedcontainers dependencies actually depend on pytest (yet!), let's hope it stays that way until the branch is merged...
<mbakke>well apart from six and importlib-metadata, but they already have the bootstrap story sorted
<janneke>oh my, to build `hello' on the Hurd, we now need guile-3.0.1
*janneke anxiously watches the build...
<GilbertVolcano>crap I'm stuck on partitioning drives, I've never done it by command line
<GilbertVolcano>Okay I figured it out :P
<Blackbeard>GilbertVolcano: do you need help?
<Blackbeard>Ah good
<GilbertVolcano>How should I partition the disk?
<GilbertVolcano>I would like to have my information on the disk encrypted.
<Blackbeard>Can you paste the result of lsblk GilbertVolcano
<mbakke>GilbertVolcano: do you intend to use the entire disk for Guix, or are you dual booting?
<GilbertVolcano>I intend to use the entire disk
<mbakke>GilbertVolcano: do you have an EFI or BIOS system?
<mbakke>if unsure, check if /sys/firmware/efi exists
<GilbertVolcano>It's EFI, but the /sys/firmware/efi path doesn't exist, I think because the motherboard is running in BIOS legacy mode.
<Blackbeard>GilbertVolcano: how did you run the USB?
<Blackbeard>My computer lets me choose between running the USB as BIOS or EFI
<GilbertVolcano>I didn't get that option, I had a bunch of issues getting the graphical installer going so I had to manually install it.
<Blackbeard>GilbertVolcano: before that, when you booted from the USB
<guix-vits>if some release years old, and much fresher git version available, and it's a game: should i attempt package a git version?
<mbakke>GilbertVolcano: there are a couple of ways to partition the disk, here are the steps I would take, assuming the disk is /dev/sdc
<GilbertVolcano>I think the disk is /dev/sda
<mbakke>right first 'parted /dev/sda' to start a "graphical" partitioning tool
<mbakke>then "mklabel gpt" to choose a GPT partitioning layout which is a little more flexible than the "legacy" MBR
<mbakke>it doesn't matter since we'll have everything on one partition, but anyway :P
<mbakke>you'll need a tiny partition to embed GRUB in, so run "mkpart" and choose a size of 1MB for the first partition
<mbakke>the start should be "0%"
<GilbertVolcano>ok
<mbakke>afterwards the "bios_grub" flag needs to be set: "set 1 bios_grub on"
<mbakke>then run mkpart again to create the main partition: start at 1MB and end on 100%
<mbakke>then you can exit parted with "quit"
<mbakke>now you should have two partitions on /dev/sda: /dev/sda1 and /dev/sda2
<gnutec>How is it the development of the shell guile to guix system?
<mbakke>next, to set up full-disk encryption, run "cryptsetup luksFormat /dev/sda2"
<TempeVolcano>let me try that now
<TempeVolcano>This is GilbertVolcano btw mbakke, I don't know why it switched to that account.
<NieDzejkob>guix-vits: if there are significant changes in the git version (and not just stuff like refactoring), then go for (g)it!
<TempeVolcano>mkpart command isn't working :P
<mbakke>TempeVolcano: what output are you getting?
<TempeVolcano>-bash: mkpart: command not found
<mbakke>TempeVolcano: ah, those commands are expected to run from within the "parted /dev/sda" shell
<TempeVolcano>I formatted sda1 with fat32, could that be the issue?
<TempeVolcano>Oh I run parted first?
<mbakke>TempeVolcano: yes
<NieDzejkob>[22:13:28] <mbakke> right first 'parted /dev/sda' to start a "graphical" partitioning tool
<TempeVolcano>got it
<TempeVolcano>got it
<mbakke>then mklabel gpt, mkpart, set 1 bios_grub on, and mkpart again to create the main partition, then quit parted
<mbakke>the file system you choose in parted does not actually matter :)
<TempeVolcano>what should I name the 1MB partition?
<TempeVolcano>GRUB?
<mbakke>TempeVolcano: sure, the name is also just cosmetic
<TempeVolcano>It's asking where the partition should start I but in "1"
<TempeVolcano>Now it's asking where it should end?
<TempeVolcano>1000? to make a MB?
<mbakke>TempeVolcano: "1MB"
<mbakke>for the other partition, start at "1MB" and end at "100%"
<TempeVolcano>Should I make it an ext4 as the file system type
<TempeVolcano>Okay I made both partitions.
<TempeVolcano>I'd like to have all this encrypted if I can.
<mbakke>TempeVolcano: the miniature GRUB partition can not be encrypted unfortunately, but everything else will
<mbakke>the purpose of that bios_grub partition is to chainload the real encrypted GRUB later on
<TempeVolcano>Okay great! It's not important to my privacy that the GRUB partition be encrypted anyway :)
<mbakke>now you should be able to run "cryptsetup luksFormat /dev/sda2"
<TempeVolcano>okay
<TempeVolcano>inside of parted?
<TempeVolcano>lol that didn't work, I guess i have to quit parted lol
<TempeVolcano>I got the error "Cannot wipe header on device /dev/sda2"
<NieDzejkob>How do I see what dependency failed here? http://ci.guix.gnu.org/build/2446517/details
<mbakke>TempeVolcano: huh, can you run 'parted /dev/sda print' and double check that it looks correct?
<mbakke>NieDzejkob: when the dependency does not show, it usually means that it does not fail any longer, i.e. has been restarted in the mean time :/
<rndd>hi people, could someone tell how to get javac?
<mbakke>rndd: install openjdk:jdk
<guix-vits>thanks NieDzejkob, i'll read their log.
<NieDzejkob>mbakke: Hmm, but search suggests that's the most recent build of audacity on master
<NieDzejkob>or do you mean the dependency? then should the build for audacity itself be restarted too?
<NieDzejkob>how do I go about making someone do that? :P
<rndd>mbakke, thank you, what is the meaning of ":" in guix packages installation?
<mbakke>rndd: it is to specify output: you can list outputs of a package with 'guix show openjdk'
<mbakke>NieDzejkob: I'll have a look on Berlin
<TempeVolcano>I'm just going to start over
<mbakke>TempeVolcano: OK, sorry for the poor instructions!
<rndd>mbakke, as i see for openjdk it is "out jdk doc", what does it mean and how to get more info?
<TempeVolcano>okay my second partition is less than 1 MB
<NieDzejkob>rndd: that means the package is split into parts named out, jdk and doc
<NieDzejkob>I'm not sure we have any documentation on what specific inputs mean
<NieDzejkob>"out" is the main, default input, "doc" usually contains documentation
<NieDzejkob>in this case, "jdk" would hold the JDK stuff, as opposed to the JRE required to run java programs
<NieDzejkob>you could use "guix build jdk" to print the /gnu/store of each output, and then run "tree" or "find" on them to look at the files each output contains
<rndd>ok, thank you NieDzejkob
<mbakke>NieDzejkob: I started a build of the audacity derivation manually and it's going now
<mbakke>not sure what the problem was :/
<NieDzejkob>Well, if it still exists, we're about to find out :D
<NieDzejkob>thanks!
<guix-vits>first thing to check "tomorrow" is if the neverball developers added `make install` :)
<NieDzejkob>tomorrow in quotes, looks like your sleep schedule has shifted :P
<guix-vits>5 AM...
<NieDzejkob>ouch
<TempeVolcano>okay mbakke got the partitions done and encrypted.
<janneke>crap FAIL: test-guild-compile
<janneke>well, same test that failed with guile-2.2, hmm
<mbakke>TempeVolcano: okay, now open the encrypted partition with 'cryptsetup luksOpen /dev/sda2 guixroot'
<mbakke>that will make it available as a normal block device at /dev/mapper/guixroot
<mbakke>then you should format it with your favourite file system, i.e. "mkfs.ext4 /dev/mapper/guixroot"
<TempeVolcano>okay
<TempeVolcano>okay done
<NieDzejkob>anybody around to take a quick look at how I added a new .scm file? https://debbugs.gnu.org/cgi/bugreport.cgi?bug=40203#17
***lx0 is now known as lxo
<ngz>NieDzejkob: It looks good.
<TempeVolcano>Whats the next step mbakke :)
<mbakke>TempeVolcano: now you need to create a config.scm that refers to the partitions you have just created
<mbakke>TempeVolcano: you should be able to figure out by reading this section of the manual closely: https://guix.gnu.org/manual/en/guix.html#Proceeding-with-the-Installation
<NieDzejkob>ngz: Thanks, I'll push the commit then
<mbakke>TempeVolcano: I forgot to mention that you should mount /dev/mapper/guixroot on /mnt before starting on that section!
<Veera>Hi guix
<drakonis>hi ho
<Veera>In pkg description how much characters per line should be there?
<Veera>hi drakonis
<brendyyn>Veera: Are you in emacs?
<Veera>no nano
<roptat>hi guix! I'm trying to configure ibus, but it doesn't save any modification, and there's something wrong with dbus it seems
<drakonis>emacs is much better equipped for this
<roptat>I get this message: dconf-WARNING **: 23:35:12.865: failed to commit changes to dconf: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name ca.desrt.dconf was not provided by any .service files
<brendyyn>oh ok. everything wraps to 80 lines i think.
<brendyyn>80 characters
<roptat>but I do have a /etc/dbus-1/system-services/ca.desrt.dconf.service
<roptat>any idea?
<guix-vits>Veera: seems nearly 74
<Veera>some pkgs are exceeding 80 characters
<Veera>Wouldn't it show ugly
<brendyyn>Veera: They wrapping is just in the source code. It doesn't come out like that in the output
<Veera>it should have been a continuous string line wrapped by editor
<Veera>But we introduce newlines ourselfs
<guix-vits>Veera: try make a long line in package and use `guix lint` -- it will complain that line longer taht X.
<Veera>guix-vits: i will try
<Veera>so guix removes the newlines introduced buy us
<Veera>and puts a space character instead
<Veera>i checked it was not showing
<TempeVolcano>Thanks mbakke, I've always been confused about mounting. What does mounting /dev/mapper/guixroot do exactly. Is /mnt the root filesystem. What does that mean exactly? I come from windows world so it's weird that the root of the file system isn't the disk you're own, it's really illogical for me for the disk to be mounted above the root of the file system. Isn't the root file on the disk?
<Veera>not showing well
<brendyyn>you can put in double new lines to separate a paragraph.
<guix-vits>good luck
<Veera>I have to try resizing my terminal characters and see the change
<brendyyn>There should be a way to do it. there is something of a format to it.
<mbakke>TempeVolcano: yes, the "target" root file system gets mounted on /mnt
<mbakke>the windows equivalent would be making the "D:" partition available on "C:\mnt" or something
<Veera>brendyyn: format is documented somewhere
<Veera>TempeVolcano: Windows gives each disk a new letter
<Veera>c: d: e: f:
<Veera>they are the root
<Veera>in linux / is root
<Veera>and every disk/ or block device is mounted in a directory
<Veera>and contents on that disk comes to be seen in that mounted directory
<Veera>the disk/block device should have valid filesystem on it
<Veera>mount actually mounts a filesystem present on the block device/disk/partition on to a directory
<brendyyn>Veera: Ok so I just checked and it's Texinfo format
<brendyyn> https://guix.gnu.org/manual/en/html_node/Synopses-and-Descriptions.html
<joshuaBPMan>Hey guix, I'm having a hard time getting my freshly librebooted latop to establish an internet connection.
<joshuaBPMan>ifconfig shows my ethernet device. enp0s25. State is up.
<joshuaBPMan>dhclient -v enp0s25 gives me an error: unable to set up timer: out of range.
<Veera>brendyyn: I remember now, I got to know basics of texinfo
<brendyyn>Veera: try putting in @* for a line break
<brendyyn> https://www.gnu.org/software/texinfo/manual/texinfo/html_node/Line-Breaks.html#Line-Breaks
<Veera>brendyyn: I will see
<brendyyn>no worries, i've never really used it much. I mostly just type in plain text ;)
<Veera>joshuaBPMan: did you configured dhcp-client-service-type in config.scm
<Veera>joshuaBPMan: Do you use one of desktops they take care of it otherwise