IRC channel logs


back to list of logs

<quiliro>i get an error when running 'sudo guix system reconfigure config.scm'
<quiliro>your guix installation is 144 days old
<quiliro>consider running 'guix pull'
<quiliro>followed by 'guix system reconfigure'
<quiliro>but it is weird because i already did that
<quiliro>what should be done? 'sudo -i guix pull'?
***danialbehzadi1 is now known as danialbehzadi
<g_bor>hello guix!
<g_bor>I have some problem I can't really find out.
<g_bor>I am trying to port guix to riscv, and while I can inherit riscv-gcc from gcc, but it seems that I can't inherit riscv-binutils from binutils.
<g_bor>gnu package base is used
<g_bor>how could i debug what is going on.
<g_bor>(it says unbound variable binutils)
<rekado_>g_bor: when porting guix, would it not be sufficient to build the bootstrap binaries and work your way up from there?
<quiliro>clacke[m]: thank you
<rekado_>g_bor: “unbound variable” can be a misleading error message. Is there any other output before that?
<quiliro>rekado_: hello :-)
<j31>hey guys, has anyone experinced trouble booting from usb after writing v0.14 i686
<quiliro>j31: i don't recall
<quiliro>j31: have you tryed?
<quiliro>j31: maybe i can guide you based on what i know, from the point where you need help
<j31>I've written the image on 3 different flash drives and none is working
<rekado_>j31: I’ve had problems with two new USB sticks (with 0.13 and 0.14); it worked without problems on an older stick (both 0.13 and 0.14).
<j31>I'll try and see if using an old usb can help
<g_bor>rekado_: It seems that no other interesting output is there.
***Guest31815 is now known as sturm
<rekado_>g_bor: cross-module inheritance is not a good idea. It can lead to top-level loops.
<rekado_>g_bor: better put the riscv-binutils in the same module as binutils itself.
<g_bor>rekado_: Yes, building the bootstrap binaries would be fine, except the standard toolchain doesn't support riscv. They practially work with a fork of the toolchain, no easy patch available, and no standalone tarballs.
<sturm>I'm attempting to package PostGIS and GDAL - but seems like the requirement json-c is missing a file called "json_object_private.h". It seems to be part of the json-c source, so any ideas why it wouldn't be coming through in the built package?
<g_bor>rekado_: I currently decided not to inherit from binutils, but what you say is worth considering.
<g_bor>rekado_: my biggest concern is, once I can build the cross toolchain the gnu build system will require additional logic to select the correct gcc,glibc,binutils...
<g_bor>rekado_: I still don't know what would be the best approach here...
<j31>hey guys, every time I write the 0.14 i686 I get this error after an fdisk -l, gpt pmbr size mismatch ..., parted automaticaly fixes this but the usb can't boot, before and after the fix
<j31>please help me out on this
<rekado_>sneek: later tell sturm The name “json_object_private.h” indicates that this is not part of the public API, which is why the build system does not install it. You may need to add the sources of json-c to the package to be able to access it.
<rekado_>sneek: botsnack
<efraim>g_bor: I thought support was being upstreamed? Are the versions in core-updates new enough?
<civodul>Hello Guix!
<efraim>gcompris-qt can't find, that's been going on for a while though
<bavier>hello Guix
<pkill9>hi bavier
<bavier>pkill9: how's it going?
<pkill9>not bad
<pkill9>how about you?
<bavier>pretty good
<efraim>Looks like my phantom file is in monolithic qt but not in qtsensors, so fixing gcompris-qt will mean fixing the modular qt build
<catonano>is it still correct to use the pypi address for python packages ?
<catonano>some tryton components can be reached both on pypi and on the tryton site
<efraim>pypi uris are prefered because we have the update checker mechanism, but if the tryton hosted ones have tests and the pypi ones don't then we'd prefer the one with tests
<catonano>efraim: ok, I'll check the tests
<catonano>efraim: thanks
<efraim>wow I have a lot of gcc tabs open, time to actually fix gcc@4.9 without causing a world-rebuild on core-updates
<quiliro> I had an error while running 'sudo guix system reconfigure config.scm --fallback'
<quiliro>I am running it again now. Is it possible to see a log of the past error?
<quiliro>Another question: On the new run, does 'reconfigure' download again the packages that it downloaded before the error?
<efraim>if it's already been downloaded it won't download again unless you tell it to
<wigust>quiliro: I believe no logs for system builds. You could redirect STDOUT and STDERR to some file, like 'YOUR-COMMAND &>/tmp/log'
<quiliro>efraim: thank you...did i tell it to download again?
<efraim>probably not
<efraim>--check would have it rebuild or redownload
<quiliro>:-) nice to know
<quiliro>is this information in the manual?
<quiliro>i will check myself...sorry
<quiliro>i am looking at section 6.2.13 Invoking ‘guix system’
<quiliro>of the guix manual
<quiliro>The manual says that 'reconfigure will not attempt to upgrade a running service
<quiliro>how can i upgrade running services too?
<quiliro>I think that the most efficient way to get and give guix support would be to program a bot to check the relevant parts of the manual regarding the question asked and print the lines and sections which are of interest
<quiliro>perhaps the bot must be very intelligent
<quiliro>if the question is not answered, then it is a good bug report for editing the manual!
<quiliro>what do you think?
<wigust>quiliro: There is no reloading services support currently, so you should stop them before 'reconfigure' to reload them during 'reconfigure'.
<str1ngs>reconfigure will still upgrade the service no? it just requires a reboot to activate the system profile?
<catonano>I was going to ask thhe exact same question
<quiliro>If the manual is correct it will not upgrade running services: "The command [reconfigure] starts system services specified in FILE that are not currently running; if a service is currently running, it does not attempt to upgrade it since this would not be possible without stopping it first."
<catonano>admittedly, it seems clear
<catonano>"it does not attempt to upgrade it"
<efraim>based on armbian's build repo, the odroid-c2 might be easier to support under guixsd than the pine64
<catonano>where would python-stdnum go ?
<quiliro>efraim: i cannot find --check on the manual
<efraim>i almost want to suggest finance
<efraim>quiliro: , search for --check
<str1ngs>seems I need to change my reconfigure workflow now :(
<quiliro>efraim: thank you...i do not have web browser now...what section of the manual is this?
<quiliro>i am using emacs infor
<efraim>quiliro: 5.1.3
<catonano>efraim: finance seems to fit, yes
<catonano>str1ngs: I reconfigured several times now while my postgresql service was running, without realizing this
<quiliro>efraim: the section which reffers to '--check' is about 'guix package'... will it apply to 'guix system' also?
<efraim>grafts throw everything off, it might, I don't think I've ever checked though
<str1ngs>catonano: ya not sure how to sanely do this now
<str1ngs>configure: error: `powerpc64le-linux' is not a supported platform. ./configure --with-courage !
<efraim>str1ngs: where'd you get a POWER system?
<str1ngs>oregon state open source lab
<str1ngs>it's part of the GCC build farm
<pkill9>ok, how do I build this:
<efraim>you'll have to get root at some point, I ran a lot of `sudo ./pre-inst-env guix-daemon ...' when I was working on aarch64
<rekado_>str1ngs: oh, exciting! Are you planning to port Guix?
<str1ngs>rekado_: I would like to, but... I don't have root access so it could be difficult
<efraim>does fakeroot or chroot require root to run?
<str1ngs>chroot does
<str1ngs>fakechroot does not
<str1ngs>my plan is to request them to install guix on a x86_64 machine. then maybe the power8 machine
<str1ngs>problem is the power8 machine is not going to be trivial. so I'm seeing how far I can port without root access first
<pkill9>if i download that into ~/guix-recipes, the run `guix build -L ~/guix-recipes chromium` it fails with
<str1ngs>pkill9: you downloaded one file?
<pkill9>yeah, i guess that's the problem
<str1ngs>if so try putting it in. ~/guix-recipes/gnu/packages
<pkill9>oh i see
<str1ngs>ideally it would be better to just clone the whole repo
<efraim>or to add it as a remote git server
<pkill9>well it finds it now, thx
<pkill9>it does have antoher error, but i think that's because i didn't clone the whole repo
<pkill9>ah now i came back to the previous error
<pkill9>i mean, the one i had before (not that it doesn't find chromium.scm)
<pkill9>it complains of bootstrap stuff and weird oddities
<pkill9>ok this is the error:
<pkill9>does that mean it is built witha newer version of guix?
<pkill9>the chromium.scm from the repo i mean
<str1ngs>I think you need to use guix from that repo
<str1ngs>not include it with -L since that will mix versions
<pkill9>oh, why would they require that heh
<str1ngs>it's not partial packages. it's complete guix as far as I can see
<str1ngs>uhhh GUILE_PROGS m4 macro is such a PITA
<efraim>ng0 keeps their WIP stuff as separate branches
<pkill9>what do they add to guix?
<pkill9>or change about guix?
<efraim>AFAIK the chromium branch just adds chromium, maybe some other changes or additions to make it build
<quiliro>efraim: --check is in the 'guix build' section...does it still apply to 'guix system'?
<efraim>quiliro: not sure, i've never tried
<efraim>it might work for 'guix system build ...'
<catonano>where would phonenumbers go ?
<pkill9>efraim: it seems to want it's own version of guix though, I wanna build chromium with the vanilla Guix
<pkill9>or something, i dont' really understand it so much, i'm probably wrong :P
<quiliro>i cannot find 'guix build system' or 'guix system build' in the manual....where is it?
<civodul>how did you search?
<quiliro>in emacs infor
<quiliro>just s? ok...thank you
<str1ngs>keeping hitting s to continue searching
<civodul>catonano: prolly python.scm
<catonano>civodul: ok, thanks
<quiliro>is pandoc included in the repos?
<efraim>if anyone wants, there's new versions of mysql and mariadb available, don't think I can get to it tonight
<quiliro>--install does not find it
<catonano>2 components of Tryton depend on each other
<quiliro>perhaps it is ghc-pandoc
<catonano>it's in the tests, for both of them. I could disable the tests for one of them
<catonano>fun :-)
<apteryx_>hello! How would I login with the ratpoison WM instead of Unity on Ubuntu 16.04?
<str1ngs>make -j150 mauuuuahahaha
<efraim>apteryx_: perhaps adding it to your .xinitrc file and then running startx from a tty
<str1ngs>alternatively you'll need a desktop file
<civodul>nckx: any idea on this btrfs-progs failure?
<civodul>or mbakke?
<civodul>the local btrfs experts :-)
<apteryx_>efraim: thanks!
<str1ngs>apteryx_: think you can copy ratpoison.desktop to /usr/share/xsessions/
<apteryx_>and should the ratpoison from Guix be installed in the system profile or as my user? I guess the later.
<civodul>the notion of "system profile" makes sense only on GuixSD
<str1ngs>yes ubuntu implies foreign distro
<apteryx_>oh, I meant 'root' profile
<civodul>on foreign distros the "system profile" is /usr :-)
<apteryx_>but I guess root profile != system profile on foreign distros?
<str1ngs>think you can do this though
<str1ngs>cp $(guix build ratpoison)/share/xsessions/ratpoison.desktop /usr/share/xsessions
<rekado_>apteryx_: that’s also the case on GuixSD.
<str1ngs>I did not test that. just in theory
<rekado_>apteryx_: the root user’s profile is not the system profile.
<apteryx_>rekado_: right. root profile is just another user profile ;)
<str1ngs>but.. with foreign disto root profile is important
<apteryx_>yeah, it contains the guix-daemon package, or so I understand.
<pkill9>hey, i dunno if it's a glitch, but `guix build` couldn't find a patch in my $GUIX_PACKAGE_PATH, so I copied it to ~/.config/guix/latest/gnu/packages/patches (I'd prefer not to as you're not supposed to modify anything in the guix store right?) and it found it. Does nayone know where I may have gone wrong?
<pkill9>the patch came from my $GUIX_PACKAGE_PATH/gnu/packages/patches
<str1ngs>pkill9: ~/.config/guix/latest is not the store
<str1ngs> /gnu/store is the store
<pkill9>it's a symlink to the store though
<pkill9>i mean, to a derivation within the store
<str1ngs>yes, but it's not thee store
<pkill9>what i meant was, by modifying a derivation that exists in the store, I'm modifying the store, and only guix is supposed to do that right?
<str1ngs>either way fix your GUIX_PACKAGE_PATH
<rekado_>how did you write there? It’s read-only on GuixSD and root-owned.
<str1ngs>rekado_: sudo! :P
<str1ngs>ACTION hides
<pkill9>rekado_: i'm running guix on a different OS
<pkill9>and yes I used sudo :D
<rekado_>yeah, don’t do that :)
<str1ngs>I'm kidding . dont' mindm e
<str1ngs>mind me*
<pkill9>rekado_: i didnt'w ant to
<pkill9>but guix didn't find the patch
<pkill9>so i thought i may as well test it
<rekado_>ACTION prepares some patches for sailfish and salmon (bioinfo tools)
<str1ngs>pkill9: paste GUIX_PACKAGE_PATH and your patch file
<pkill9>also, it did find my package recipe in my $GUIX_PACKAGE_PATH
<str1ngs>pkill9: to pastebin
<str1ngs>GUIX_PACKAGE_PATH you can paste here
<rekado_>if the patch is directly in GUIX_PACKAGE_PATH then it should be found.
<pkill9>my GUIX_PACKAGE_PATH is /home/itsalsome/guix-recipes
<rekado_>if it was in $GUIX_PACKAGE_PATH/gnu/packages/patches then it won’t be found
<pkill9>ah ok
<pkill9>that's where it was
<rekado_>that’s because $GUIX_PACKAGE_PATH/gnu/packages/patches is not among the default patch paths.
<pkill9>oh ok
<str1ngs>is patches special?
<pkill9>so i just put patches in /home/itsalsome/guix-recipes in this case?
<rekado_>str1ngs: the point is that it is *not* special.
<rekado_>that’s why the location $GUIX_PACKAGE_PATH/gnu/packages/patches isn’t looked at
<rekado_>it’s just in the Guix sources that $srcdir/gnu/packages/patches is defined as a directory to look for patches.
<rekado_>this does not extend to GUIX_PACKAGE_PATH
<str1ngs>ah guix sources
<str1ngs>I just use GUIX_PACKAGE_PATH was not aware of patches in guix sources
<str1ngs>err source*
<str1ngs>what should a git commit message look like for a new package? for sending to
<str1ngs>I've packaged python2-libtorrent and deluge
<rekado_>Are you using Emacs? We have Yasnippet templates for common commit messages.
<rekado_>but you can also see the format in git log
<str1ngs>I use emacs, but not yasnippet but I can set it up. if it makes my contribution workflow streamlined
<str1ngs>I also need to setup emacs/git for email patches
<str1ngs>I guess I can do 'gnu: Add python2-libtorrent, deluge'
<rekado_>they should be separate commits.
<rekado_>for setting up the snippets take a look at doc/contributing.texi
<rekado_>the snippets work with magit.
<str1ngs>ok, I'll try to split this commit
<str1ngs>I can just fiddle with the hunks in magit though
<quiliro>it is almost time to leave...but 'sudo guix system config.scm --fallback' is not through yet after 12 hours
<rekado_>this command does not look right.
<rekado_>Also: be sure that the root user’s guix is up-to-date.
<apteryx_>Copying the ratpoison.desktop file didn't enable switching to it from lightdm
<apteryx_>xinitrc work though. maybe I should disable lightdm and always do "startx" from the terminal at boot... I don't boot often anyway.
<quiliro>rekado_: it said it was 141 days old
<str1ngs>apteryx1: did you restart lightdm?
<quiliro>should i run 'sudo -E guix pull'?
<str1ngs>sudo -i guix pull
<quiliro>str1ngs: thanks
<quiliro>after that 'sudo guix system reconfigure config.scm --fallback'?
<quiliro>i get errors with 'sudo -i guix pull'
<quiliro>ice-9/boot-9.scm:760:25: In procedure dispatch-exception:
<quiliro>ice-9/boot-9.scm:760:25: In procedure struct_vtable: Wrong type argument in position 1 (expecting struct): #<pointer 0x0>
<quiliro>i will report in
<str1ngs>quiliro: can you do guix pull as normal user?