<thomassgn>I'm getting "/etc/config.scm:97:19: Wrong type to apply: #<service-type cups" when trying to configure cups in my guixsd config. Line 97 is: "(cups-service-type" anyone have a clue? <thomassgn>neither the services/cups.scm or the docs help me understand this one <mekeor>does guix compare hashes to verify packages? what about irreproducable builds? are guix packages deterministic? <bavier>mekeor: we try to make packages as reproducible as possible <bavier>mekeor: we have 'guix build --rounds=...' and 'guix challenge' to help track determinism <quiliro>i have problems with updating guix from 11 to 12 <quiliro>some error regarding mariadb and InnoDB ***erdic_ is now known as erdic
***samis is now known as CompanionCube
<rekado>thomassgn: could you show us your config.scm? ***jonsger1 is now known as jonsger
***katco` is now known as katco
<slyfox>what is the difference between master branch and core-updates? <snape>slyfox: core-updates is for things that would trigger a large number of builds <snape>the point is for hydra to have time to build them, before they got pushed to master <slyfox>does core-updates get regular master merges or only when something is actually pushed to core-updates? <snape>I don't know the merge process <lfam>slyfox: Master is not merged automatically into core-updates. We do it manually <lfam>I don't think it could be done automatically. There are usually merge conflicts that require human review <slyfox>right. basically core-updates is not meant for users at all <lfam>slyfox: Well... I hope that some of us try to use while it's still being built, so that we can catch bugs before we merge it back in to master. But for most of the core-updates cycle, important things may not build at all <lfam>At the beginning of the cycle, we are just updating core packages on that branch without testing if they build or not. It's computationally too expensive to use at that point, because every day a full rebuild may be required. <lfam>We haven't been sticking to it as closely as I'd like. But, it takes time and energy to shepherd the branches along. <mthl>lfam: Did you keep this as a bookmark in you web browser? :) <lfam>Even better, if I type "branching" into the address bar that URL is the first thing that shows up ;) <lfam>slyfox: I'm not that familiar with the process of adding support for new architectures. But in general, if some change is useful on its own and does not cause too many rebuilds, it can go to master. Otherwise, to core-updates. <lfam>efraim may have some advice if he is around. He's been doing the aarch64 port <slyfox>AFAIU it changes glibc and triggers rebuild for everything <lfam>Indeed, that's a core-updates change <lfam>But, I'm not able to review it since I don't really understand it. Somebody else will have to do that :) <kyamashita>Do most of our perl packages have time-dependent reproducibility issues? I'm encountering problems with .pod file generation. <slyfox>'guix build --check a-package' should catch it, right? <kyamashita>slyfox: It should, and it is in this case. I'm looking over Marius' package series right now, and that seems to be the blocking issue. <lfam>And you can use diffoscope to inspect the differences <slyfox>kyamashita: out of curiosity which package is it? <lfam>My understanding is that issue should have been fixed in the last Perl update. Their release notes seemed to imply that. But I've heard others complain that it's still happening :( <kyamashita>slyfox: The one I've investigated so far is perl-async-interrupt (to be commited). <kyamashita>The difference is literally one timestamp in a .pod file. <kyamashita>On my system, "perl -V" gives "Jan 1 1970 00:00:00" like it should. <kyamashita>As I go through the diffs between each build, I only see the timestamp and the resulting hash of the package change. If the problem is with perl and that's the only problem these packages have with reproducibility, I'll continue my evaluation normally. <lfam>Makes sense. We should fix the issue with Perl <quiliro>is that correct for searching for package that has iperf command? <quiliro>for other cases where it is not the name of the package but only the name of the command, would it work? <kyamashita>quiliro: I don't think so. That reminds me of the "apt-file" utility in apt-based distros. <Apteryx>quiliro: Per the manual, what "guix search" does is: "List the available packages whose name, synopsis, or description matches REGEXP." <jmi2k__>I can't install GuixSD with LUKS, can anyone help me? <jmi2k__>At boot, I get the following message, and I get a guile prompt: <jmi2k__>ERROR: LUKS partition not found #vu8(232 32 171 206 34 211 64 255 164 158 255 145 146 140 50 221) <kyamashita>jmi2k__: Huh. This error is new to me. You setup your partitions according to the Guix manual? <jmi2k_>kyamashita: yes I think. Well, root has no label, but it shouldn't cause any problem as I'm using /dev/mapper/root <sneek>From what I understand, core-updates is fontconfig, eudev and cups-minimal FTBFS, tcsh's test hangs, util-linux needs a fix for CVE-2017-2616, gnutls lacks IDN support unless it uses libidn2 <jmi2k_>Also in mapped-devices, the source field uses the LUKS uuid instead of a path, but that is fine according to the documentation <jmi2k_>kyamashita: yes, with minor differencies. The UUID is defined as %root-crypt-uuid, which should make no difference. Also, device field is "/dev/mapper/root" and the title field is 'device. The rest looks the same. <jmi2k_>kyamashita: the target field is "root" <adfeno>CharlieBrown: Generally, Emacs can interact with your desktops paste natively. <jmi2k_>I'm installing GuixSD on a usb drive, could it be the problem? <adfeno>CharlieBrown: Althought, if you are not uisng GuixSD, you are probably like me: Still figuring out how to make Emacs indeed interact nicely with your clipboard. <adfeno>Well, "PasteBin" itself isn't one, specially because it requires visitors to run non-free JS. <adfeno>ix.io could be an interesting service provider, but I don't know how to make Emacs interact with it. <adfeno>paste.lisp.org is also interesting one, but I say the same as I did with ix.io. <kyamashita>jmi2k_: Heh, reading over the manual, your config is more correct than mine. It might be the fact that you're installing to a USB drive, though I'm not sure. <jmi2k_>I can't think of another reason. I've had this problem before, but I gave up and installed it without LUKS. I think I'll do the same this time... <CharlieBrown>adfeno: DejaVu Sans Mono lets me distinguish those chars. GNU FreeMono is very thin and hard to see. <adfeno>How do I check which font I'm using? <jmi2k_>kyamashita: Yes it works. In fact, I've been carrying it in my USB drive for three months without any problem <jmi2k_>That's why I take care of using UUIDs whenever possible <adfeno>CharlieBrown: Hm... What is the result of the following: M-x describe-font RET RET <kyamashita>jmi2k_: Strange that LUKS breaks booting from your USB drive. I'll have to try this on my own USB hard drive and see if I can reproduce this. <jmi2k_>kyamashita: ok thanks. I'll keep trying more things <adfeno>CharlieBrown: Try actually installing the dejavu fonts. <CharlieBrown>Another thing that the native IceCat package does that the Guix one doesn't: middle click + trackpoint to scroll; pretty icon provided in launcher. <adfeno>CharlieBrown: For pretty icon, one must do the following: <CharlieBrown>Oh yeah, and the scrollbar in Parabola IceCat actually responds to my Numix theme. <adfeno>ln -s "../../../.guix-profile/share/applications/icecat.desktop" <adfeno>This will get the IceCat icon in the menu/launcher. <alezost>CharlieBrown: what do you mean by "raw"? <alezost>When you call "M-x debpaste-paste-region", it pastes the selected region and puts the paste URL into clipboard. <alezost>When you call "M-x debpaste-display-paste", it receives the paste from the URL at point ***jonsger1 is now known as jonsger
<ng0>*grummel* what am I missing in offloading when on the build-machine I have (guix config) and can load it fine with guile, but the 'guix offload test' throws me guix offload: error: failed to use Guix module on '192.168.1.179' (test returned #<unspecified>) ? when I run the ssh command to execute the guile command, the error says that the module doesn't exists, yet the user profile on the build-machine has it <ng0>been months since I've had offloading <ng0>just 'guix package -i guix' in the profile? that would be weird. I followed the instructions <osmano807>Anyone experiencing 504 errors with hydra.gnu.org? <osmano807>I was having issues with mirror.hydra.gnu.org too <osmano807>I switched my substitutes to mirror.guixsd.org and my guix started working again <osmano807>Maybe there's some network strangeness going on <kyamashita>Maybe. I get 404 error sometimes, but it's usually my wireless network acting up. <quiliro>i cannot find pdf editor on the repositories <quiliro>Apteryx: so what would be the way to search for the package that has a certain command...like the "apt-file" utility does as kyamashita says <roelj>How can I set the mount options in /etc/fstab in GuixSD? <roelj>Mount options for the root filesystem.. (I can't even find them in there) <rekado>quiliro: for PDF annotations you can use xournal. 