<davidl>unmatched-paren: Is there any way to use the search-input-file procedure when you are using the trivial build system? I was trying to make everything work with the new style there before switching to the copy build system.
<davidl>For the other build systems, people use (lambda* (#:key inputs #:allow-other-keys) to get the inputs variable which the trivial one doesnt have.
<dcunit3d>i've got it to remap with a loadkeys script, but after i press the capslock for control, it's like the control modkey gets stuck on. it ends up affecting all of the sessions, instead of the active one, so i have to restart my laptop
<ulfvonbelow>hey question, what should happen in (gnu build file-systems) in 'mount-file-system' if 'mount-may-fail?' is #t but 'check?' is also #t?
<jpoiret>civodul: wdyt of regenerating bootstrap blobs for the Hurd?
<jpoiret>i'm thinking the least invasive approach to fix coreutils booting would be to add the kernel headers to the bootstrap blobs. Also, I want to try whether the m4-boot0 succeeds with newer headers, but I'm not sure it will help or not
<jpoiret>for some reason Guile just loops if I try to use %bootstrap-coreutils&co instead of coreutils-boot0 in gcc-boot0's source
<civodul>jpoiret: why would they need to be regenerated?
<ulfvonbelow>background: I've got a raid array, and each drive should contain a copy of the grub-efi bootloader, so booting is always possible. So those all have their own efi system partition, and an attempt should be made to mount each one, but it shouldn't break things if one of them fails. If non-risky repairs and checks can be done at the time, that would be good too. But of course if the device doesn't exist, it shouldn't fail.
<efraim>civodul: what do we have inside the snippet, just guile and explicitly imported modules? No packages?
<ulfvonbelow>currently when I unplug a drive between boots it finishes activation and then hangs forever
<jpoiret>civodul: we're missing the kernel headers for coreutils-boot0, and I wanted to see if newer headers would make the m4-boot0 tests crashing go away
<jpoiret>efraim: well, I'm not trying to remove the snippet, only remove the dependency on coreutils-boot0 which doesn't build
<jpoiret>i was thinking of using %bootstrap-coreutils&co but guix doesn't like that for some reason, gets into an infinite loop
<bumble[m]>no recent mentions here about sound suddenly stopping-working, but in case this helps anyone... my sound suddenly stopped working a week or two ago. I opened pavucontrol tonight and found the audio muted (while playing a video in qutebrowser; video needs to be playing for the muted device to appear in pavucontrol) after un-muting the device bam sound is working everywhere (yay)
<jpoiret>apteryx: I also get etc/teams.scm complaining when I use `mumi send-email` with patches which i guess just uses git send-email under the hood.
<dcunit3d>if an invocation of `guix build` is failing when using the meson build system, how do i get to the see what was building? particularly, the meson-log.txt?
<dcunit3d>i can see the build log at *.drv.bz2 and the last derivation. i'm trying --keep-failed
<dcunit3d>ok, nvm i forgot the --keep-failed on the last run. i think most of what i need is in the /tmp folder
<trankitron>FYI I've already solved the problem with encrypted root partition. The problem was the password I was introducing was wrong because the first time the password is required the keyboard layout still is not changed to the configured value (to ES in my case).
<jpoiret>ah yes, that's a common pitfall unfortunately
<jpoiret>janneke: yes, i still had most of the work done in a git stash from when I was trying to make hurd boot
<dcunit3d>is it possible that i need to build it with a smaller -j flag?
<dcunit3d>it's a meson-build-system package though
<jpoiret>dcunit3d: you'll need to provide some more logs. But yeah, since ninja is highly parallelized, it can happen that it launches a bunch of linking jobs at the same time and that ends up eating all your RAM
<ennoausberlin95>chomwitt: You are right. I checked the guix source tree and there is no (gnu home services ssh-agent). What do you want to configure? BTW: Scheme-Fu is weak in me, so I might be wrong
<ennoausberlin95>chomwitt. You can see the details if you clone the guix source and check /guix/gnu/home/services/ssh.scm. There is a record-type and services defined for the agent configuration, but the defaults are reasonable
<bjc>is bordeaux ok? i keep getting 404 errors from it. lately it's the ‘which-2.21’ nar
<dcunit3d>for shepherd-0.10.0 should we import guile-fibers into the system packages?
<dcunit3d>or is that a bad idea? i think the build for shepherd bundles in everything it needs anyways
<janneke>ACTION again forgot that something like this doesn't work yet:
<janneke>ACTION goes to add a cross-built procedure to map over inputs
<jpoiret>janneke: I have a much better solution! create a file manifest.scm, and in it, use (guix profiles) and (gnu packages hurd), and add `(package->development-manifest hurd #:target "i586-pc-gnu")`
<jpoiret>yes, this exists and seems to do exactly what you expect :)
<jpoiret>then you can `guix shell -m manifest.scm` of course
<lfam>formbi: Works for me, finishes in a couple seconds
<efraim>formbi: maybe it's related to the commit you've pulled to? what commit do you have from `guix describe`?
<chomwitt>ACTION thanks <ennoausberlin95> for looking
<tschilp>formbi: by what you describe I did have the same thing with e499cb2c12d7f1c6d. 26b71e1982afac91 did not have the memory problem any more, but errored out as well. Yesterday evening's c5bc698e8922d78 brought things to normal for me.
<janneke>ACTION changes #:make-flags into an upstream patch for the Hurd
<tschilp>formbi: right now's 1638d1f0cecb52a7392d78534e9a0136878759e4 does fine here -- so just running a 'guix pull' before running the 'guix pull --news --details' might solve the problem. The commit you mention is exactly the one, that blew my RAM yesterday too. Thanks to efraim, who saved my brains on that :)
<tschilp>What's the difference between linux-libre and linux-libre-bpf? 'guix show' gives me the same description for both...
<jpoiret>tschilp: I guess the second one has BPF enabled?
<jpoiret>would be nice to have that information in the description though, yeah
<jpoiret>if you don't know what that is, you probably don't need it
<tschilp>jpoiret: and now I have to look up what bpf is, though I most likely don't need it ;)
<formbi>tschilp: ah, maybe the newer one doesn't have a substitute available (I use channel-with-substitutes-available)
<formbi>as for the bpf - it's very annoying when people make a package variant and not put anything new at all in the description
<formbi>but well, that's ok for the holier-than-thou maintainers
<tschilp>formbi: might be -- for my system and home configs everything was there fortunately! Actually still true for the last one I mentioned above (or better -- none of the packages I use got updated since yesterday night, they seem very brand new to me anyway).
<lfam>The lack of package description was surely not intended to annoy anyone or be "holier than thou". It was surely a simple oversight that could be easily corrected
<jpoiret>formbi: maintainers are not the only ones who push commits by the way :)
<jpoiret>and of course, anyone can help review patches!
<formbi>jpoiret: I mean, they sometimes block people's submissions using weird excuses
<formbi>jpoiret: read the whole thing, my favorite quote is this: «I found myself often changing my mind one way or the other quite often back then, seems like currently it's back to opposal on basis of licensing.»
<tschilp>lfam: now that's what I call quick! Cool!
<PotentialUser-32>tschilp Yeah I know, but it did work some times if I recall correctly. I'm running the guix system.
<jpoiret>formbi: ah, so it's not possible to change opinions over a period of time?
<ieure>dcunit3d, Not sure what you mean by that. Jails like... FreeBSD Jails?
<dcunit3d>ieure: i thought the ZFS file system was integrated with BSD jails or maybe it's the other way around.
<ieure>dcunit3d, No idea. I run it on Linux, the good stuff is _really_ good and the bad stuff is mostly non-critically bad but very annoying.
<ieure>Bad stuff I've run into are like, snapshots and replicating datasets.
<ieure>There are multiple things that try to do this, they're all hideously overgrown very fragile shell scripts. I gave up and set up Syncthing to replicate a subset of my stuff offsite and that worked immediately and with way less horseshit.
<dcunit3d>my exposure to OpenZFS was limited to what I was doing with proxmox. From how the manual indicated the features that it offered (when using proxmox), I used it, but I dreaded any issues that I encountered that would require the cmdline.
<PotentialUser-32>tschilp what do you mean by checkout? If you mean I'm learning and experimenting on my own, kinda, I have someone that helps me but I don't like to trouble him too much. I'm still reading the manual but some stuf is still not easy. I think my problem is linked to the fact that I have a channel from which I pull packages from, and for some reason
<PotentialUser-32>when I reconfigure there is something that he doesn't like. Since it asks for the auth token, I guess I need to put the token I use somewhere, probably in the config.scm I'm thinking, but I'm not sure how yet. there is also a problem with the icecat browser, but I'll probably just uninstall that since I never use it.
<formbi>PotentialUser-32: coule you send your config.scm?
<tschilp>PotentialUser-32: I thought you are maybe building guix from a local clone of the repo, I every now and then play with that, but also had some authentication problems (but that was local and gpg, and a misunderstanding on my side). As far as I know git authentication is not supported, so you might be better off putting your channel into a public repo, or if it's public anyway, you could try to use the https instead git url.