IRC channel logs
2024-08-15.log
back to list of logs
<zeropoint>lispmacs: you might be interested in the vm-image.tmpl file in that same directory <nikolar>i have a question, i noticed that libical wasn't being built for aarch64 because rust wasn't built <nikolar>but the only missing dependency is gtk-doc <nikolar>wouldn't it be better to split the package into the actual package and the documentation so that the package itself doesn't depend on gtk-doc and inkscape and rust <graywolf>Yeah, it seems *really* useful, but cannot figure out how to get rid of the errors/warnings :) <rgarcia>hi! will guix rebuild all kinds of stuff if I just want to add a user? <jaft_r>rgarcia: I think it'll depend on if there are updates that you've pulled down. If you haven't ran "guix pull" (or have but there weren't any commits pulled in that would trigger an update in any of your installed packages), then you want. But (I'm assuming you're adding the user via your config.scm) anything that'd trigger a rebuild of your system profile will occur during a reconfigure. <rgarcia>jaft_r: thanks! I realize I did a guix pull, so triggered updates. <weary-traveler>i believe efraim is on the team, but i believe he's currently travelling. expected to be back around september <jaume>Hi! I am building a package for 'network-manager-strongswan'. This is the first time I build a package, so maybe I just need a stupid and quick advice. The package builds fine, but when it enters in the 'install' phase it fails because of permissions. My guess is that it tries to write on NetorkManager package directory. This is the error: https://paste.sr.ht/~devvesa/1c814b783adda0f725be102398db9b1bba833bb6 <jaume>any help super appreciatted! <Rutherther>yes, it seems from the outupt that it located network-manager under there, and tries to write it there. Look up how the package finds out this directory and pass correct one (being "(assoc-ref outputs "out")"or via gexp "#$output"), ideally the package should take it in env var, if it just assumes it's always under network manager, then you might need to apply a patch <civodul>ACTION made good progress fixing i586-gnu (Hurd) native builds on ‘core-updates’ <attila_lendvai>hi civodul, any word on the shepherd patches i've sent to the list? they are straightforward ones, just factoring out repeated patterns (and with that introducing a placeholder for potential asserts and logging). <civodul>attila_lendvai: sorry no, not yet; i need to allocate time for that (still on vacation…) <attila_lendvai>civodul, i've looked at your recent commits an i have to tell you that some of them are recorded almost identical in my branch... it sure feels like wasted effort (regardless of the ordering). <attila_lendvai>(deprecation warnings, service printer, to pick two recent ones) <civodul>attila_lendvai: oh, i understand and i’m really sorry about that <civodul>i have to admit i’m not really able to keep up <rmin>Hello! A question to aid my mental model of guix: Can we create a full iso for a specific operating system configuration that doesn't need to compile or download substitutes, to use it as an offline usbstick image? <AwesomeAdam54321>The log rotation only happens occasionally, rather than when a log file exceeds a given size <Rutherther>rmin: isos are built with "guix system image". You give it an "operating-system" record, so you can choose packages, services that go onto it. Is that what you are asking? <AwesomeAdam54321>There's the risk of the disk space being fully utilised by very verbose logs before the next log rotation occurs <rmin>Rutherther: This will bundle all binaries with the iso? So no download is needed first using it on a new computer <Rutherther>rmin: it will build the whole operating system, yes. For starting the computer and using what you specified in your config you won't need to download anything. <fnat>What's the difference between 'guix system image --image-type=docker config.scm' and 'guix pack --format=docker package-0 package-1 ...'? I have some intuition, with the former you're able to define "system things" such as services as opposed to only package dependencies? Is that correct? <fnat>Brilliant, thanks AwesomeAdam54321. :) <civodul>AwesomeAdam54321: yes, good point, it’s modeled after old-style cron + rottlog <civodul>i agree we should add the ability to have a size threshold, that makes more sense <civodul>actually there’s already one, so it’s possible to check more frequently and with a higher threshold <civodul>we’d need something more sophisticated to fully avoid periodic checks <attila_lendvai>does hibernate work with full disk encryption? if yes, how do i trigger hibernate from the command line to test it? <dariqq>attila_lendvai: i have set it up on a laptop with fde and a swapfile and it was working last time i checked (though i have not ben using it lately). Im using 'loginctl hibernate' but iirc i had to force a hibernation 'manually' once for that to work <zberkowitz>Are build logs returned by --log-file supposed to contain the full config/build/test text? I'm able to package some things well enough, but when I want to look at the logs with --log-file I just see some lines describing the grafts <attila_lendvai>echo disk > /sys/power/state, and there's `loginctl hibernate`, but it fails with "Not enough swap space for hibernation" (i have RAM/2 swap) <ieure>attila_lendvai, Hibernate saves the contents of RAM to swap, so you need at least as much swap as physical RAM for it to work. <attila_lendvai>ieure, it's compressing the ram, and it only needs the currently used amount, so... that's not true as is. <attila_lendvai>loginctl has some mindless check that can be disabled with some env variable (don't get me started) <ieure>attila_lendvai, Yes, you can use less if you want, if you're okay with it just failing to work sometimes, depending on how much stuff's in RAM and how compressible it is. <attila_lendvai>now that was strange: it seemingly hibernated, but then it came back alive. hrm... maybe it was my kinesis keyboard. it does interfere with suspend-to-ram. <zberkowitz>AwesomeAdam54321: OK, thanks. I'll give that a shot here in a sec <ieure>Hibernate to a LUKS2 swap partition works on Debian, though Debian has an unencrypted /boot (which Guix doesn't). Guix seems to prefer a swap file, at least, the graphical installer doesn't configure a swap partition and it's irksome to do so, since you also have to recreate the root partition. Not sure what the bounds of the problem are in this case. <attila_lendvai>strange. it seems to get into hibernate, but then the power button blinks 3 times, and it comes back alive. there's nothing obviously interesting in /var/log/messages, just the logs of hibernation. turning off hw, etc. and my IRC connection survives it, even though the network card is reinitialized. <dariqq>ieure: wouldnt resuming from an unencrypted swapfile/partition defeat the purpose of disk encryption? <rynn>Ahh, good to know, I have 64gb of ram but only a 4gb swap, so... <ieure>dariqq, The swap partition on Debian is encrypted, /boot is not. <attila_lendvai>yeah, the main point of suspend-to-disk is that it's way much more secure. a suspend-to-ram laptop can be exploited with physical access through who knows how many ways... including the freezing and grabbing the ram modules. <ieure>attila_lendvai, IRC connection surviving doesn't surprise me at all, the network was down for a short time, so any missed packets got retransmitted. Idle TCP connections can survive a laptop going into sleep mode. <ieure>Only idle ones, though. If the other end tries to send packets and gets no ACKs back, it'll close on its end. <ieure>I pretty regularly have SSH sessions survive the laptop being closed and asleep overnight. <rmin>Has somebody shared their guix config or can point out a great one to learn from? <pi`>Has someone shared their guix config or can point out other good ones for learning? (Sorry for asking again, disconnected) <civodul>i had put you in X-Debbugs-Cc but mumi or something overrode that <ryan77627>Hi all, question regarding debbugs since I am new to email based git workflows. If I want to test a patchset that exists in debbugs (72457), is there a quick way to actually download all the patches from the list other than doing it manually? Especially since I don't actually have any of these emails in my actual inbox, so I can't just apply the patches from there. <ieure>ryan77627, There's a mumi invocation that'll do that; it's listed at the bottom of every issue on issues.guix.gnu.org. <ieure>ryan77627, There's also some thing somewhere that applies patch series to Git repos, but I cannot find it. <ieure>I continue to dislike both the email patch flow and proliferation of tooling needed to patch over its deficiencies. <ryan77627>ieure: Thanks, I've actually never used mumi so I figured it was more specific to interacting with the web interface. Running into a new issue now, just a git merge conflict, which I think(?) I can handle. Though, is there any way to limit what mumi is trying to merge? there have been afew revisions at this point and I'd like to just merge in the v3 patchset, but it looks like it is just starting <ryan77627>Or, rather, does it just do this automatically? <ieure>ryan77627, I honestly have no idea, because I've never used mumi. <ieure>Just, have heard it mentioned & people seem to be into it. <ryan77627>I shall keep looking around then, in that case. Thanks for the help so far <Franciman>is there any tutorial on how to setup emacs for guix? <Rutherther>I am not sure about a tutorial. What do you need to know though? <Franciman>what packages would be helpful. Like paredit? <Franciman>or is there something more sophisticated nowadays for structurally editing s-exps? <Franciman>maybe this is a question to better ask in #scheme <Rutherther>Ah, maybe yeah. Personally I havent gotten into using those packages much yet, though Ive heard geiser is good. <freakingpenguin>ryan77627: I haven't use mumi, but there's an http endpoint you can pull specific revisions from. issues.guix.gnu.org/issue/<issue_num>/patch-set/<revision_num>. <janneke>civodul: oh my, hurd was broken again on core-updates <freakingpenguin>Huh, I can emulated build a aarch64 system itself, but when I try to deploy to it Guix tries building qemu-minimal and that build fails. <freakingpenguin>i.e. $ guix system build os.scm --system=aarch64-linux works, but $ guix deploy machine.scm where machine.scm contains the previous os for aarch64-linux fails. <weary-traveler>ieure: is there a forge that allows commenting in-line on changesets via email? <rgarcia`>hi! while trying any subcommand I get "guix install: error: error parsing derivation `/gnu/store/772cziqznpnmadgkm0gp03l9bgkkdp". That .drv file is empty. Help! <Rutherther>have you tried "guix gc --verify="contents,repair""? It might run for a while. It's for repairing a corrupted store. <duncan>anyone using tcsh (or something whicgh isn't compatible with bourne shell) as user shell? display managers seem unable to launch sessions after setting it <duncan>non-specific errors like 'gdm: Gdm: GdmDisplay: Session never registered, failing' <duncan>I did pput all the GUIX_PROFILE (&c.) --everything in /etc/profile in my cshrc and in /etc/csh.login, and they are there if I log in via tty <rgarcia`>hum, this contents,repair command is finding a bunch of empty drv files. I wonder. <Rutherther>what fs are you using? have you had any power outages? <rgarcia`>no power outages that I can recall. But something weird did happen: it is an old laptop with no battery for the bios so the date is wiped upon power cycle. Erroneusly i set up a future date when I installed the system. So my first generation was/will be from october! <rgarcia`>do you think I should do a fresh install? <Rutherther>if repairing will fix it, then it's fine. If it doesn't, then you might have to do something like that. <rgarcia`>*sigh* no luck. Well, still having fun. Thanks Rutherther!