<pkill9>am i supposed to install xf86-video-nouveau to use the nouveau driver ?
<vagrantcish>so, trying to package a u-boot variant not in mainline u-boot ... i point the source to the git url, and everything goes fine, except it fails to build, seemingly missing a bunch of include files ...
<vagrantcish>but when i do "guix environment --container" and run the commands manually, everything works fine.
<apteryx`>Hello! Why is my guix-daemon using /usr/local/etc/guix instead of /etc/guix, and how can I fix this?
<vagrantcish>what would be present in "guix environment --container PACKAGE" that wouldn't be with "guix build PACKAGE" ?
<apteryx`>nckx: But if I used the official 'guix pull' to deploy versions of guix, shouldn't that override whatever I could ever have done when playing with the guix source code? And a 'guix pull'd guix should use /etc/ out of the box, no?
<apteryx`>also, how do I get the manuals for linux system calls available on GuixSD?
<nckx>apteryx`: Perhaps you're running a different daemon than you think? I can't help with ‘guix pull’ since I've never really used it.
<nckx>(bootloader (bootloader-configuration (bootloader grub-efi-bootloader) (target "/boot/efi"))) has always worked flawlessly for me.
<nckx>It sounds like your EFI system partition isn't mounted or not at the point where GRUB expects it to be.
<nckx>Is /boot/efi mounted? Are you still in the installation environment? Recent installers have a bug where they expect the EFI system partition to be mounted at /boot/efi, not /mnt/boot/efi which would make more sense.
<RetardedOnion>i have my /dev/sdb1 (which is the $esp) mounted at /mnt/boot/efi. i am booted in the install env.
<RetardedOnion>so umount esp and mount at /boot/efi(which should not be empty?
<RetardedOnion>i expect issues. its beta and i am an arch user. i cant help programming but i will be helping at the software availability. i think the whole system is a great start at packagemanagement which does not suck. i wonder why i didnt find it earlier
<pkill9>you can specify one in the config but that's obviously a security risk
<RetardedOnion>i think i just got the beta experience... well it could be asked when you guix system init /mnt/etc/config.scm /mnt
<nckx>Guix has its fair share of suck, like all software, but it's a promising attempt at managing the suck.
<RetardedOnion>i think i expressed that already. also: why no networkmanager for xfce?
<nckx>“The ‘root’ password in the new system is initially empty; other users’ passwords need to be initialized by running the ‘passwd’ command as ‘root’, unless your configuration specifies otherwise (*note user account passwords: user-account-password.).”
<nckx>If you missed that or think it could be improved, please shoot an e-mail to the list.
<RetardedOnion>oh yeah. mail lists. generally the installation could be very much more simple for new users with a small ncurses interface... though it is easy if you understand the concept
<nckx>There's an unfinished installer but I don't think it's seeing much development at the moment.
<very>I have a quick question concerning program packaging under Guix.
<very>I'll soon be releasing a program that uses some libraries I've also written. I'm using a sh program to launch it and I'd prefer it to be the same across GNU/Linux systems; it works relative to the user's home directory. My question is if Guix permits packages that add a directory to the home directory, when it's necessary.
<very>This is the most natural fit, with what I want to do with the libraries, which I'd prefer be separate packages.
<apteryx`>very: many programs already add stuff to a user's $HOME directory, such as Icecat (.mozilla) and countless others.
<apteryx`>so I don't see why your program shouldn't be allowed to.
<very>So, there shouldn't be an issue, then; I just thought I'd make certain.
<snape>civodul: here https://gnunet.org/bot/log/guix/2018-06-18#T1738636 you told me why hydra failed to build guix-modular. I said I understood but in fact I don't. What are those two different guix modules you are talking about? To me there is only one: the one that is passed as an input.
<snape>It's probably too obvious, but to me everything being compiled comes from the --fresh-auto-compile option in hydra-eval-guile-jobs.in
<snape>but if it's the case, then I don't understand why it would work with other kinds of jobsets
<pkill9>it also makes it so it doesn't re-download files if they've already been downloaded when it runs `guix import` :3
<Copenhagen_Bram>well when i cloned it from github, there was a checkbox to determine whether or not it was a mirror. I wasn't sure what it meant. What's the difference between cloning a repo and cloning it as a 'mirror'? But I checked the box. Now you're telling me it's not official because it's a mirror. So maybe I could clone it again and make it not a mirror.
<snape>Copenhagen_Bram: what checkbox are you talking about? I've never seen such checkbox
<Copenhagen_Bram>farewell for now guix, i must shut down my wifi to conserve internet. I have a partially downloaded Dragon Ball Z Abridged playlist to entertain me until my internet is fast and free once more
<mbakke>civodul: Thanks! Quasiquoting the (add-after ...) block was what I needed :-)
<jlicht>how could I adapt the recently packaged maven to play nice with java 8 based projects?
<mbakke>Why does `guix build -d -s i686-linux openblas` require downloading a bunch of i686 substitutes?
<mbakke>`guix build -d -s i686-linux gcc` returns just the derivation as expected.
<civodul>mbakke: because of grafts; try --no-grafts
<mbakke>civodul: Aaaahhhh of course. --no-grafts works, thanks :-)
<ProfessorDey>okay, so I slightly fucked up. I was trying to get my system image install to regenerate properly, forgot that gc existed and stupidly did a find | rm -r for the packages I wanted gone. Now everything is obviously broken. Is there a way to regenerate the manifest or at least rebuild everything currently listed as installed?
<janneke>g_bor: do you know what this means: ld: access beyond end of merged section (1728814200)
<janneke>is my .o file fu, is ld broken, or could it be a version mismatch (gcc-2.95.3 + binutils-2.25) and should i try+debug binutils-2.14 with gcc-2.95.3?
<roptat>ProfessorDey: do you still have a working guix command? what did you remove exactly?
<ProfessorDey>roptat: I was removing a number of packages that were created using `guix system disk-image`, because for some reason it was ignoring the changes I made to the initrd.
<roptat>ProfessorDey: so you removed items from the store?
<heobjo>I found out that the new guix pull also has generations, i.e. list them with the -l option. But there is no --help on how to deal with them otherwise than showing. Can I switch back to one, or do anything else cool with them?
<Sonolin>can anybody answer if GuixSD will always stick with Shepherd as init (PID 1)?
<Sonolin>trying to find a FOSS system without SystemD (and don't really want to try Dragora)
<Sonolin>bavier cool yea elogind was what I was thinking, glad to hear its already included!
<Sonolin>maybe I'll actually try gnome-shell - last time was with Ubuntu when they first switched to Unity, been without a DE since :P
<bavier>I've used gnome only occasionally, for testing mostly, but it seems to work nicely
<ng0>rekado: you've done some fair amount of work with Sun units, right? I'm not sure if you ever dealt with the pricetags on them, but an almost fully equipped Fire X4450 unit for less than 300€ is that worth it?
<pkill9>but i mean sometimes you have to add '--prefix=' in 'configure-flags' in the package definition, if it's adding that automatically then why does it sometimes need to be added in the package definition?
<bavier>oh, I see what you mean; do you have an example package?
<rekado>but one could specify a new package variant and refer to that instead of the variant that is commonly used.
<Sonolin>oh ok, then I'm assuming the variant could be installed alongside the main (latest) version?
<rekado>the result is a new library variant under its own prefix; the application using it is configured at build time to use just that variant, and it embeds a reference to the absolute file name of that library variant.