IRC channel logs


back to list of logs

<CornBurglar>I posted earlier, sorry if I missed the answer due to leaving, but has anyone heard of successfully putting guix on OpenBSD?
<civodul>CornBurglar: that's not possible
<civodul>Guix requires the kernel Linux or the Hurd
<civodul>wigust: is it your first upgrade to the new 'guix pull'?
<CornBurglar>alright, thanks
<wigust>civodul: no, guix --version => 8edebd3207335aff15081458e895104ac9f45d3d
<wigust>‘guix pull -l’ shows 4 generations
<wigust>i found that ./pre-inst-env should be invoked inside ‘guix environment --pure guix --ad-hoc help2man guile-sqlite3’ in 31841
<tang0>civodul: thank you for moving me in the right direction. The installation finished without errors. However, the new system does not boot.
<tang0>I also noticed in the installation wiki where it *specifically* says to mount the ESP partition on /mnt/boot/efi or guix init will not find it at boot time.
<tang0>but doing so, makes the install fail every time.
<civodul>tang0: hmm we'd have to discuss how it fails to boot
<civodul>i'm leaving for today, but perhaps you could email help-guix all the details
<tang0>civodul: I'll do that. Thank you for all the help. I appreciate it.
<Copenhagen_Bram>hey guix, why wont libgme compile?
<mbakke>Copenhagen_Bram: It looks like an instance of <> :/
<lgen>What do i have to do to add a different output from a package with multiple outputs to the packages field of the operating system declaration?
<reepca>hm, when I try running "guix system reconfigure /etc/config.scm" on my laptop, I get some issue about not being able to find a command named system...
<reepca>I also seem to be having some issues with the networking being inconsistent, which is strange (works for awhile after restarting the networking service, then stops working)
<apteryx`>about that missing system command, I had that, it's caused by a bump in upgrading guix to the new guix pull IIRC.
<apteryx`>Try installing guile-sqlite3, I think that is what is missing
<apteryx`>lgen: I seem to recall its documented in the manual
<apteryx`>(list package
<apteryx`>output)’ syntax can be used to add a specific output of a package
<apteryx`>in Guix info manual, node 6.2.1 Using the Configuration Syste
<OriansJ>question, what is the preferred way to have the linux libre bzImage in /boot so that all of the system (minus /boot) is fully encrypted?
<mbakke>OriansJ: AFAIK Guix does not have a facility for placing the kernel other places than the store. So you'd have to write some custom code.
<efraim>Custom service?
<vagrantcish>i'm trying to add python-libusb1, but it uses ctypes.util.find_library to find ... and fails to find it ... there doesn't appear to be a way to pass the library path to the find_library python function ... how is this sort of thing typically handled?
<vagrantcish>since libusb is passed as an input, a particular build should have that input hard-coded somehow?
<vagrantcish>it looks like i can patch it dynamically
<brendyn>Copenhagen_Bram: Did you manage to finish your package definition?
<snape>OriansJ: but it's fine for linux to be encrypted
<snape>it's actually more secure
<g_bor>Hello guix!
<g_bor>Is there any way to ask guix build to retain the build directory if the build does not fail?
<g_bor>I'm trying to debug a build, and of course I can make the build fail, but if there is an easier way, that would be great.
<demotri>g_bor: I once had the same issue. Other than failing with #f/an exception, I'm not aware of a special flag like -K or something.
<janneke>g_bor: i'm not aware of that feature, would be nice to have
<janneke>apropos, it would be *so nice* if upon build failure, you could drop into a shell with the exact environment settings, and if all sources would have been put in a local .git archive after unpack
<demotri>janneke: But you know that you can do that manually: guix environment <package>; cd /tmp/guix-build-<package..>; source ./environment-variables.
<demotri>janneke: And the source code: ./pre-inst-env guix build -S. OK, it's a tarball, not git.
<g_bor>demotri, janneke: Thanks, I've ended up throwing an exception. It would be nice to have, as it would make debugging no fatal build failures easier.
<janneke>demotri: unless the package does `(setenv ...)...and adding the source to git is pretty difficult after stuff has been built
<janneke>demotri: hmm, a package could add a `populate-git' stage
<janneke>during development this could be a feature
<demotri>janneke: Why do you want to add to git anyway? I would expect the code to be checked in already somewhere.
<janneke>demotri: `want' is possibly a bit too strong, i'm brainstorming
<janneke>demotri: i'm working on GuixSD bootsrap build right now, and must somehow manage, write and rewrite patches for a number of versions and packages
<janneke>i find myself using a shadow build infrastructure, using a card-house of shell scripts
<janneke>it would be great if changes could be automagically handled by git, the environment by guix, and a couple of edits and rebuilds could be done in place
<demotri>This looks like a mes to me :-) I'm sometimes reading your announcements. Looks great, but hard to understand all the details for me.
<janneke>demotri: yeah, mes :-) I'm hoping that I find a way to make it easier to understand once the project starts to land in GuixSD
<brendyn>What's the default --cores value for guix pull
<efraim>It's hardcoded to build guix with just a single thread
<efraim>But all the other parts use all the threads allowed to guix-daemon
<brendyn>Ok it failed when i set --cores=9 and left some guile process using 100% of a thread. maybe i shouldnt try that again
<brendyn>is guix supposed to always be build with 1 thread?
<jlicht>hey guix
<janneke>hey jlicht
<sahithi-ihtihas>Hello Guix
<jlicht>and hello sahithi-ihtihas
<janneke>hey sahithi-ihtihas
<sahithi-ihtihas>I have added colorize-string procedure and soft port to guix/ui.scm earlier and i got colorized strings but when i try to do it now i am not getting colorful outputs and soft port is not being called
<Copenhagen_Bram>brendyn: no i didn't
<sahithi-ihtihas>All changes are in on my branch...
<pmikkelsen>hi guix
<Copenhagen_Bram>hey pmikkelsen
<OriansJ>snape: that is true but it requires GRUB_ENABLE_CRYPTODISK=y and knowing where to hook that in the process is unforunately what I was hoping someone else would know.
<snape>OriansJ: I don't know how to do it without Libreboot, but if you happen to run Libreboot, I can help.
<OriansJ>snape: sure, I'll take a look at that
<pkill9_>does kxmlgui fail to build for nayone elese?
<snape>pkill9_: yes, it looks like the Qt upgrade issue
<snape>probably a matter of adding the missing header
<mbakke>OriansJ: We do set GRUB_ENABLE_CRYPTODISK=y. Encrypted /boot is supported for both EFI and BIOS GRUB.
<rekado>the gtk-doc update broke libosinfo, and thus gnome.
<rekado>“import six” fails.
<rekado>looks like it needs to propagate python-six
<rekado>oh, well, it *is* propagated. Hmm.
<snape>does changing Guix's code while build operations are being done affect them?
<snape>I guess it doesn't
<snape>Specifically, what I want is: start a build, change code, start another build, change code, start a third build
<snape>the builds take a lot of time
<rekado>I’ll wrap the gtk-doc executables
<pkill9_>i don't think it affects them snape, i would have thought it would load the file into memory, and then read that file in the memory
<pkill9_>but i could be wrong
<apteryx`>hello! is this the right way to use 'guix copy'? guix copy --from=x220 /gnu/store/gpxvy2zmgbaz2jcf1szcbsm294ygwxsn-googletest-1.8.0
<apteryx`>it recognizes my host, but says: retrieving 0 store items from ''...
<apteryx`>while I'm sure the above store URI exists on that x220 machine.
<apteryx`>nevermind, it already was there ;)
<mbakke>rekado: Can we also remove the six propagation from gtk-doc now?
<efraim>i'm up to "checking if the VM and compiler work together... configure: error: VM failed to run compiled class." for icedtea@1 for armhf-linux
<vagrantc>so, i'm trying to update python-trezor, and it requires adding python-libusb1 and python-pyblake2, which i've managed to get working, but i haven't yet gotten the test suite for python-trezor to work
<vagrantc>does it make sense to submit patches for python-pyblake2 and python-libusb1 if nothing else uses them, or should i wait till i have working patches to update python-trezor?
<vagrantc>oh, this is also sort of needed to upgrade electrum... which i forgot is where i started on this journey :)
<vagrantc>well, trezor support in newer versions of electrum is broken without an updated python-trezor
<Copenhagen_Bram>Hello guix! Is qt 4 available for guix?
<g_bor>vagrantc: I guess it does make sense to submit the patches. They will be needed, when you manage to get trezor working, and if someone else is working on packages that need them can save some duplicate work.
<vagrantc>g_bor: makes sense
<vagrantc>is there an option with "guix environment FOO" that actually also downloads and unpacks the source tree for FOO and drops you into the source directory?
<Copenhagen_Bram>So, I've finally got rid of all the errors in my qtads package. Now I need to fix compiling errors. It looks as though there is no ./configure. So how do I make it run qmake instead?
<g_bor>Copenhagen_Bram: yes, we have qt 4.8.7.
<Copenhagen_Bram>how do I add that as an input
<g_bor>vagrantc: I don't think so...
<efraim>Copenhagen_Bram: ("qt-4" ,qt-4)
<vagrantc>ACTION ponders writing another feature request bug :)
<efraim>not unexpected, but definately a setback, copy-paste arm(v7?) assembly for use in aarch64 failed horribly for jamvm@1.5.1
<vagrantc>heh :)
<Copenhagen_Bram>alright, now how do i modify the build system to use qmake?
<snape>Copenhagen_Bram: you replace the 'configure' phase with a phase that calls qmake, I believe
<snape>Copenhagen_Bram: see the 'librecad' package for example
<Copenhagen_Bram>ACTION closes his ears to the corrupt church of emacs propaganda
<Copenhagen_Bram>oops wrong channel
<vagrantcish>ok, so i've got python-libusb1: ... i've pretty much cargo-culted parts from python-pyusb, so if anyone sees anything obviously wrong or unecessary ... otherwise it seems ready for submission
<Copenhagen_Bram>ACTION looks at librecad and becomes confused
<Copenhagen_Bram>that's a lot of modifications
<j3kyl_>hey, how do I run jar files under guix, cant find openjdk or whatever
<demotri>j3kyl: icedtea is the build-wrapper for openjdk.
<demotri>j3kyl_: guix package -i icedtea. Then just as usual: java -jar myjar.jar
<civodul>hey there!
<civodul>does java-plexus-container-default ring a bell?
<civodul>actually that's not the main problem here
<mbakke>Huh, that patch was removed some time ago.
<civodul>yeah something's wrong on
***jonsger1 is now known as jonsger
***jonsger1 is now known as jonsger
<reepca>Hm, I'm in a bit of a pickle right now on my laptop... installed onto a fresh hard drive using the 0.14 installer, ended up with root having guix 972b87b14ae54bacf2457a4d538c5495bec37176 and now guix pull produces an error message about "./guix/build/compile.scm:132:21: ./guix/build/compile.scm:132:21: In procedure memq: Wrong type argument in position 2 (expecting list): (#<intset 6209> . #<intset 6410,6420,6430>)". And I'm still
<reepca>getting an error message when I try using "guix system" (guix: system: command not found), although I've tried installing guile-sqlite3 as someone recommended.
<apteryx`>did I read recently that our default database choice was mariadb? (over mysql or sqlite?)
<apteryx`>But the thread about the guile-sqlite3 dependency being required to unlock 'guix system' was here:
<apteryx`>reepca: ^^
<apteryx`>IIRC, after I had guile-sqlite3, 'guix system' magically became available.
<apteryx`>that means you'd have to install it for your root user before attempting a guix system reconfigure, I believe.
<apteryx`>but it seems you have a bigger issue ATM (guix pull failing with the error you mentioned).
<reepca>apteryx`: root user has guile-sqlite3 in its profile, at least as far as guix package -I is concerned.
<apteryx`>OK. And still no 'guix system'?
<apteryx`>does it have guile as well?
<apteryx`>basically you need to install all the dependencies of guix until you can successfully 'guix pull' into the new era.
<apteryx`>I think someone posted a 'guix environment guix --ad-hoc guile-sqlite3 ...' command somewhere which is supposed to achieve this.
<reepca>so wait, "guix pull" depends on stuff being installed in the user's profile? That seems a bit strange
<apteryx`>this changed recently, but the 'guix pull' on this way is a bumpy road.
<apteryx`>recently, Ludo introduced a Guix which comes with all its dependencies (IIUC).
<rekado>reepca: “guix pull” does not depend on stuff being installed in the user’s profile.
<rekado>the error with “guix system” disappears when running “guix pull” again.
<rekado>if you’re using Guix from a git checkout, though, it does require that guile-sqlite3 is present in the current environment.
<reepca>I'm just using what's globally-installed
<rekado>I don’t understand how this relates to what I wrote.
<rekado>“guix pull” does not install anything globally.
<rekado>note that when you use “guix pull” it installs Guix into ~/.config/guix/current/bin/guix, so you should use that variant of Guix going forward.
<reepca>I suppose what I meant was that the guix program being used is under /run/current-system, and not in a git checkout
<reepca>(although currently I'm trying to see what happens if I install guix for root and run it from /root/.guix-profile/bin, and for some reason pulling seems to be getting a lot farther there)