IRC channel logs

2017-02-19.log

back to list of logs

<ng0>3 hours past. It's safe to assume that the python package testsuite just hangs, right
<ng0>my feeling..
<catonano_>ng0: I would easily assume that, yes
<ng0>one would assume that the build environment notices after a while
<Guest51825>dang, I seem to be stuck not being able to install anything due to #25775. If anyone has managed to recover from this, i'd love to hear
<Guest51825> https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25775
<emyles>'guix offload test' for a newly setup host gives this error:
<emyles>failed to use Guix module on 'pluto.noogony.co.uk' (test returned #<unspecified>)
<emyles>What might #<unspecified> mean?
<emyles>Guest51825: I think I just sent you a message? was it private? wasn't meant to be.
<Guest51825>thanks emyles, sorry to hear you had to reinstall
<emyles>I'll have go at fixing things for the sake of progress but sometimes re-installing is the only sane thing to do. It doesn't take too long.
<emyles>Here's a strange thing:
<emyles>guix package -I
<emyles>Is it version 2.0.12 or 2.0.13?
<Guest51825>emyles: so did the problem occur again when you did a `guix pull`?
<thomassgn>how do I choose version for packages? like guile, https://www.gnu.org/software/guix/packages/g.html says there is guile version 1 and 2, but their names are the same.
***jonsger1 is now known as jonsger
<Apteryx>I have a list of regexps, and a list of strings to string-match with those regexps. What construct should I use? So far I have something like a for-each and map/string-match... but there might be something more tailored for this use case?
<Apteryx>thomassgn: guile@1
<Apteryx>as in, "guix package -i guile@1" You can be as specific as you want with the version following the @.
<Apteryx>sneek, later tell emyles: the <unspecified> means exactly this: the value is unspecified. Example: (unspecified? (if #f 'never)) --> #t because the "false" branch of the "if" is unspecified.
<sneek>Will do.
<Apteryx>seek: botsnack
<Apteryx>sneek: botsnack
<sneek>:)
<Apteryx>I'm thinking when I have nested map in an outer map it's probably a case where I should fold.
<Apteryx>ACTION reads the Guile manual
<Apteryx>Do we have a mechanism to print/log messages from a service definition?
<Apteryx>Here's my map in a map; there might be a nicer way to do this? http://paste.lisp.org/display/339464
<janneke>Apteryx: you want to use filter
<Apteryx>janneke: Makes sense now that you say it :)
<Apteryx>Thanks!
<Apteryx>I guess this will also remove the need to use the ugly (remove unspecified? ...)
<janneke>Apteryx: exactly
<rekado_>I get this error when trying to use “guix system container config.scm”: guix system: error: service 'file-system-/sys/fs/cgroup/elogind' requires 'file-system-/sys/fs/cgroup', which is not provided by any service
<catonano_>I have an important announcement: I managed to install GuixSD on my laptop !!
<rekado_>catonano_: woo!
<catonano_>now it doesn t let me log in as a normal user AND I can t reproduce the root password. But I m happy anyway :-)
<catonano_>I remember thomasd discussing a bug for which when logging in as a normal user it would retake you to the login screen
<ng0>if you did n't set a root password, it's empty
<ng0>you have to set it in tty
<catonano_>ng0: it was empty. I set it
<janneke>catonano_: yay!
<catonano_>janneke: :-)
<catonano_>I had messed up the file systems layout, I think, that s why he installation didn t work correctly
<janneke>catonano_: you had to overcome some obstacles and frustrations, no?
<janneke>ah
<catonano_>I think I will reinstall it now and leave the root password empty until my emotions settle
<catonano_>I remeber thomasd discussed a bug for which when logging in as a normal user, it took you again to the login screen
<janneke>catonano_: you are using more than 1 partition?
<catonano_>I m running into it right now
<catonano_>yes I am using /boot, / and /home
<catonano_>janneke: do you remeber what the solution was ?
<janneke>catonano_: no, i don't remember this
<janneke>ACTION did not have this problem
<catonano_>I ll write on the help mailing list
<janneke>catonano_: i'm just using /, any good reason not to do that?
<catonano_>janneke: no, none good reason. Just habit. I like to think that should I want to reinstall, say, Fedora or GuixSD from scratch, I could keep my home folder
<catonano_>it s just, you know, a thought I m used to have
<janneke>catonano_: i see. my first installation of GuixSD used 30GB for /
<janneke>twice as much as I was `used to use' for /
<janneke>that proved waaaay to small when I decided to work on Guix's cross-build feature
<catonano_>I gave 100G to / it happened in the past that I couldn t build wiredtiger because there was no more room on / and I wanted to make sure not to run into that again :-)
<janneke>catonano_: i'm just terribly frustrated by how easy it is to share /home between different distro's (read: used to be fine, now next to impossible?)
<catonano_>janneke: next to impossible ? Why ?
<janneke>but maybe i'm just too stupid to setup this
<janneke>catonano_: it seems anything gnome just doesn't work, dunno
<catonano_>janneke: ah maybe different versions of Gnome use different file formats for preferences
<janneke>file-names, paths into to /usr or /gnu/store, dunno?
<janneke>catonano_: possibly also that
<catonano_>or maybe they store preferences in different versions of those desktop services storing prefs on the disk
<ng0>with the kgpe-d16, the only problem one might encounter is ram issues, right? that's basically what the Guix server ran into I mean..?
<janneke>i used to have sort-of-a declaritive way to setup gnome, using gnome-settings
<janneke>but it seems that things change
<catonano_>janneke: anyway, you can keep your firefox bookmarks, your pictures, your PDFs
<catonano_>that, to me, has some value
<janneke>:-)
<janneke>catonano_: sorry for the rant, if you have some success/recipe to share desktops between distros, i'd like to know ;-)
<catonano_>and should I come back to normal mail from Gmail, also email
<catonano_>janneke: no worries ! I was just discussing :-)
<janneke>:-) glad you're on GuixSD!
<catonano_>janneke: yes, it s awesome :-)
<catonano_>ng0: what is kgpe ? I m not sure I understand what you re talking about
<ng0> https://www.gnu.org/software/guix/news/growing-our-build-farm.html
<catonano_>ng0: ah ok. I thought you were talking about my installation. Sorry ;-)
<rekado_>ng0: there are many more problems one might encounter.
<rekado_>ng0: the board has a couple of limitations when used with libreboot.
<rekado_>some of these limitations are unknown because the number of configurations that have been used with libreboot is small.
<ng0>ok
<rekado_>I’m really confused by the guix-devel “discussion” about bitmessage
<rekado_>is there some back story?
<rekado_>I’d like our communication on guix-devel to be supportive and friendly. I think this hasn’t happened in that case.
<ng0>there is a backstory.
<ng0>nothing I can point to, but dealing straight forward with Peter is the best you can do
<rekado_>there are a lot of terse, cryptic comments there that seem a lot like someone is angry for some reason and only makes the least possible effort at getting understood.
<ng0>if I wouldn't have done this, i wasn't angry, it would've gone on forever
<rekado_>my preference is to not engage rather than to communicate poorly.
<ng0>I keeped getting mail in my inbox for no reason..
<ng0>next time I simply ignore
<ng0>are you interested in the backstory?
<rekado_>no
<ng0>ok
<rekado_>I think discussions on guix-devel should be self-contained.
<rekado_>I’m getting way more mail in my inbox than I can handle. I don’t think this is a good reason to communicate in riddles.
<ng0>I'd say it's typical for Peter that this happened..
<rekado_>please don’t make this personal
<rekado_>as far as guix-devel is concerned it doesn’t matter.
<ng0>I don't, the backstory isn't even personal. But I'll go back to other things now
<rekado_>ok
<ng0>regarding the thread on help-guix where I was "asked", I won't respond to that person. If someone doesn't understand why, ask me off-list.
<ugoday>Hi there. Does xfontsel work for you?
<ugoday>when I launch it I got err https://paste.pound-python.org/show/lk38M3yeg0R7Dg8bJlqv/
<janneke>ugoday: yes, works for me
<janneke>ugoday: what font packages do you have installed?
<ugoday>45 packages with name like 'font-'
<ugoday>I've noticed that I don't have xset.
<jmd>ugoday: xset has its own package.
<ugoday>yep, with xset it works. Is there any easy way to do xset +fp /path/to/fonts for all fonts already installed?
<jmd>Well the fonts should be in the profile. So I suppose xset +fp ~/.guix-profile/wherever should work.
***jonsger1 is now known as jonsger
<efraim>Looks my test failure in coreutils on aarch64 is known upstream and patched
<efraim>I can probably work around it with parallel-tests? #f
<katco>i'm receiving a backtrace/error when attempting to perform various options with guix. this began after performing a pull sometime last week. couldn't find any reference to this on the mailing lists; can anyone help me get back to a good version?
<Apteryx>katco: I'm not sure how to recover from guix failures when using "guix pull" (if ever this command fails). What I know would work is to checkout guix from git, build it there, and use it using the ./pre-inst-env script.
<Apteryx>if even*
<efraim>It could also be an old version of the daemon
<Apteryx>efraim: I'm still unclear as to what action is supposed to update the daemon? Is "guix pull" responsible for this? What about using guix from git?
<katco>Apteryx: hm... i thought about that, or just fixing the error in the source file to un-stuck me, but wasn't sure if there was a less kludgey practice
<katco>here's a paste of the issue: http://pastebin.com/b9UbRHLN
<efraim>With guix on Debian, I made sure the daemon was pointing to /var/guix/per-user/root/bin/guix-daemon or something like that
<efraim>And I have guix installed as root and keep that one up to date
<efraim>That backtrace looks like the other too-old guix-daemon ones
<katco>efraim: that may be it... systemd is telling me the daemon is coming from a 0.10.0 dir
<efraim>Yeah, that one is pretty old now
<katco>efraim: ok, so... i'll try and update that. curious... how is that supposed to be updated? manually by me?
<efraim>Im actually on my phone so I can't check exactly what I have on mine
<efraim>I think since then we've tried to change it so that it would point to Root's install of guix
<efraim>And then you just have to keep root's copy of guix updated
<katco>efraim: ah yes, ran into "too many levels of symbolic links" issue with those instructions
<Apteryx>efraim: OK. I think I remember now. What version of the guix-daemon runs depends on what reference to it the service definition has, right?
<efraim>Right
<Apteryx>I'm trying to run an instance of my config.scm in a vm (using the script generated from "guix system vm /my/config.scm", and it hangs at "Logging in..." screen. Anything I should try?
<Apteryx>Nevermind, that static logging in image was misleading.
<Apteryx>The system works, I could start a terminal and type stuff in there :)
<katco>efraim: hm, well i repointed systemd to a 0.11.0-1.4420 daemon, but still getting a backtrace. looks like same backtrace =/
<Apteryx>katco: what does "guix build guix" return? I'll try to see if there were any fix to packages.scm file recently.
<katco>Apteryx: http://pastebin.com/2XWsQbeH
<Apteryx>I know that I'm running guix-daemon 0.12.0 (git checkout), so it could be that 0.11.0 is still too old.
<katco>hrm that was the latest one i saw on my system
<katco>find /gnu/store -name guix-daemon
<Apteryx>katco: I think you want "find -L /gnu-store -name '*guix-daemon*'"
<Apteryx>/gnu/store *
<katco>Apteryx: thanks. 11 still latest version found =/
<katco>Apteryx: unless 12 doesn't have the version in the path as 10 and 11 do
<Apteryx>Your find command seems sufficient. Sorry for the noise.
<katco>Apteryx: no worries at all. you're attempting to help for while i am grateful!
<Apteryx>katco: guix-daemon here shows under: /gnu/store/9hhljacc22jppmjx57xc7c46by10y8gh-guix-0.12.0-4.d9da/bin/guix-daemon
<katco>Apteryx: yeah hm don't have that =/
<Apteryx>Guix is not very hard/long to build locally if you want to try to.
<katco>Apteryx: i'm not ultimately opposed, but i was trying hard to stick with the "as a user" experience
<Apteryx>katco: OK :)
<Apteryx>Sounds like a worthwhile thing to try. You probably want to open a bug ticket to track its resolution/document it.
<katco>Apteryx: i'm going to work under the assumption that 11 is still too old and see if i can't side-load 12 into the store somehow
<katco>Apteryx: yeah should investigate opening a bug ticket too
<Apteryx>katco: Actually, your bug seems already opened here (assuming it can be fixed by using guix-daemon 0.12.0): https://debbugs.gnu.org/cgi/bugreport.cgi?bug=25775
<katco>Apteryx: oh, cool! i was looking on guix's mailing lists. running ubuntu as well
<katco>Apteryx: according to this, it looks like the 12 daemon will fix it :)
<katco>Apteryx: now to figure out how to side-load 12 haha
***jonsger1 is now known as jonsger
<Apteryx>katco: Apparently guix-daemon is part of guix package. So you need a guix-0.12.0 package in your store and reference its bin/guix-daemon executable.
<Apteryx>Actually if you are manually testing abstain from touching the store directly at all.
<Apteryx>Just leave the guix-daemon somewhere else (i.e. where you built in or extracted it) and point your service to it.
<katco>Apteryx: k
<Apteryx>I'd expect such an archive to be found in the guix 0.12.0 official binary release.
<katco>Apteryx: so, build it manually, point systemd to the out-of-band bin, guix pull, repoint to official 12?
<Apteryx>katco: Hopefully this should work!
<katco>Apteryx: i think it should!
<Apteryx>katco: Not sure if systemd supports symlinks location, but if it does you could simply point it to whichever profile is expected to have the most up-to-date guix-daemon executable.
<katco>Apteryx: i'm not a systemd expert, but i think it does. however, i tried to set up a new guix install last weekend and ran into a "too many levels of indirection" issue with symlinks
<Apteryx>OK
<Apteryx>I know upstart doesn't which is annoying.
<Apteryx>Any way to use dvorak as a keyboard layout with qemu's -k option?
<katco>Apteryx: daemon expects 12 under the /gnu/store path... wonder if i'd mess things up terribly if i just copied the 12 installation onto my existing /gnu/store
<Apteryx>katco: Hmm. Create a backup at first to be on the safe side.
<katco>Apteryx: yeah. i mean it probably shouldn't since the store is *designed* to isolate versions
<Apteryx>My thinking. But if your store state is valuable/expensive to recreate, backing it up is wise anyhow ;)
<katco>Apteryx: haha true enough. recreating wouldn't be so bad tbh
<jmd>But the store is useless without the database. So you should back that up too.
<Apteryx>jmd: Database being in /var/... ?
<jmd>Apteryx: Usually, yes.
<Apteryx>OK. In my case I have a complete rsync backup of / so I should be covered ;)
<jmd>Personally I just backup /etc and /home
<Apteryx>OK. I might go that route after I'm completely satisfied with my /etc/config.scm.
<jackknife>Is it possible to compile programs locally? like seperate from guix
<jackknife>I have the most specific issue atm. Everything works surprisingly well on my laptop but apparently some guix magic goes on when it comes to lisp. It ends up using an old ASDF version and basically makes quicklisp worthless
<jackknife>So I'm trying to figure out if I need to compile it locally with its seperate asdf version or some make an editted version of the gnu build-asdf thing
<ng0>I think I was wrong about eventlet being a part of python
<ng0>*python3
<ng0>debian sid packages it for python3
<ng0>and there is a separate repository on github which does not state that it is a part of core
<ng0>any pythonists which see it differently? Otherwise I'll just package that and hope it fixes the pykka test for me
<Apteryx>ng0: let me check
<Apteryx>ng0: What makes you believe it should be part of Python3? A search through Python3 docs for "eventlet" returns nothing, which suggests it's not.
<Apteryx> (https://docs.python.org/3/search.html?q=eventlet&check_keywords=yes&area=default)
<ng0>some bad line in a file I've read
<ng0>ok
<Apteryx>Can I use (display "some message") in a service definition?
<Apteryx>I guess it would work but wouldn't be very useful unless shepherd pipes the stdout to some log file.
<ng0>yar! fixed it
<Apteryx>good! :)
<ng0>yeah and I'm back to the issue I asked this week about.. why was this stupid git fix applied? basic commands are fixed, but send-email has been broken ever since for me
<ng0> https://lists.gnu.org/archive/html/guix-devel/2017-02/msg00804.html
<ndowens08>Hmm does guixsd not support uefi?
<Apteryx>ng0: Hmm. Just trying git send-email.
<Apteryx>ng0: I don't have send-email installed for older git I'm using (guix package -i git@2.11.0:send-email says: "guix package: error: git: package not found for version 2.11.0".
<ng0>the issue is in the commit I gave with the git after that
<ng0>not in an older one
<Apteryx>Yes, I just wanted to validate that it worked in the first place.
<Apteryx>But I guess if you tell me it did.
<Apteryx>:)
<Apteryx>Let me try the newest git then.
<ng0>it used to work
<ng0>I know because I use it in scripts
<Apteryx>OK.
<Apteryx>ng0: Why did you unset GIT_EXEC_PATH?
<Apteryx>It works when it's set.
<ng0>because it broke even more
<Apteryx>OK. At least "git send-email" works after I eval'd "export GIT_EXEC_PATH="/home/maxim/.guix-profile/libexec/git-core${GIT_EXEC_PATH:+:}$GIT_EXEC_PATH"
<ng0>i think the path I got might've introduced errors for me. I wasn#t able to use git anymore wit hthe git exec path exported
<ndowens08>Search seems to say that guix doesnt support uefi, any idea when it MIGHT be, estimate, not expecting a exact date or so
<jonsger>oh nice, network works on qemu :)
<Apteryx>jonsger: with the "-net user" argument?
<emyles>ayt sneek?
<sneek>emyles, you have 1 message.
<sneek>emyles, Apteryx says: the <unspecified> means exactly this: the value is unspecified. Example: (unspecified? (if #f 'never)) --> #t because the "false" branch of the "if" is unspecified.
<jonsger>Apteryx: no, with model=rtl8139 and I switched to LAN instead of wlan
<emyles>Apteryx: thanks, I looked into it but couldn't see the problem so asked on the ML
<Apteryx>No problem :)
<Apteryx>jonsger: OK! Can you ping using this scheme? Or is ICMP something which can't work in QEMU for some reason.
<jonsger>I'm not sure but now I think I forget to use dhclient :P
<Apteryx>Q: What's the recommend approach to raise an exception in Guix? Or print a warning message? Any pointer welcome.
<Apteryx>*recommended
<ng0>jonsger: I'm interested in the asnwer about icmp aswell
<ng0>i was told it is not possible, but the person trying to help must've forgotten it
<ng0>so I keep testing on bare-metal, which makes testing slower
<ng0>push every change set, pull, reconfigure,..
<jonsger>Apteryx: what do you mean with "ping using this scheme"? ping 8.8.8.8?
<Apteryx>jonsger: I read in the Guix manual that pinging doesn't work due to ICMP (7.2.14 Running GuixSD in a Virtual Machine).
<jonsger>Apteryx: yes pinging doesn't work for me here. I tested always with "guix package -i hello" which throw errors if there is no internet connection
<Apteryx>jonsger: OK! Makes sense.
<jonsger>emacs-minimal 40MB. lol
<ng0>spice/usbredir is gone
<Apteryx>Would anyone know how to restablish the "C-c . u" shortcut/minor-mode that was loaded in Emacs by default before?
<ng0>got the right url. I will send an update
<ndowens08>Can guixsd be installed on a UEFI system? Manual doesnt mention it
<rekado_>Apteryx: that’s guix-devel-mode, which is no longer part of Guix. You have to install “emacs-guix” because it’s now a separate package.
<Apteryx>rekado_: Thanks. I guess (add-hook 'scheme-mode-hook #'guix-devel-mode) will do in my ~/.emacs
<Apteryx>ndowens08: I don't have a UEFI system myself, but IIRC civodul might have one.
<Apteryx>ndowens08: So I would expect it to be possible, although it might not be well documented yet.
<ndowens08>Apteryx: ah, well i have nixos on this system, thought about removing arch and putting guixsd on that partition
<Apteryx>ndowens08: I think it must be as simple as telling the grub configuration of you config.scm to use "grub-efi" instead of the "grub" package.
<ndowens08>Apteryx: I dont even use grub, but refind
<Apteryx>ndowens08: Looks like you'd have some guix packaging to do at first then (we don't seem to have a refind package available in guix).
<rekado_>Apteryx: I have this in my init.el: (add-hook 'scheme-mode-hook 'guix-devel-mode)
<rekado_>my whole init file can we seen here: <https://git.elephly.net/?p=software/emacs.git;a=blob_plain;f=init.org;h=2d895b4e4a4293c0fa051fdc3e97b361f5d414d4;hb=HEAD>
<rekado_>it’s one big org document