<acrow>Yes, rekado is an example case in real life as well as the documentation.
<trevdev>If I did something silly like installed php composer via just downloading composure.phar with the copy-build-system and moved it directly to #$output/bin/composer, and that was successful, how do I get to execute it without root? I'm getting permission denied.
<trevdev>Of course the composer.phar has to do something weird all unto itself to verify its own integrity. I imagine the patching of the shebang changed its SHA512 fingerprint and now won't run because the signature cannot be verified. Cool!
<trevdev>Oh well, back to just moving the thing to ~/.local/bin/ for now. Maybe later
<Gooberpatrol66>Is it unusual for guix to restrict /run/dbus? A program i'm trying to run is trying to access it but my user account isn't allowed to access it
<Gooberpatrol66>my gentoo machine lets me read /run/dbus. how does guix handle dbus?
<acrow>rekado: system reconfig completed successfully. It took awhile to build the new rust dependencies but, eventually, alacritty rebuilt successfully!
<kiasoc5>i cant run qt5ct/other qt apps on wayland, even tho qt-wayland and qt5ct are in the same profile
<zamfofex>rekado: When trying to offload builds to my childhurd, it complains that ‘chroot’ isn’t available. Which is fair enough, but I see no way to skip that check.
<zamfofex>More acxcurately, I get an error like this: ‘derivation '/gnu/store/…-module-import-compiled.drv' offloaded to 'localhost' failed: chroot builds are not supported on this platform’
<apteryx>you can setup your daemon to disable chroot I think
<apteryx>pass "--disable-chroot" as an extra-options to your guix-configuration
<zamfofex>I see. 🤔 Is there any way to disable the check without having to disable native chroot builds on my host?
<zamfofex>apteryx: I’ve added ‘--diable-chroot’ as an extra option to my operating system definition’s ‘guix-configuration’, ‘sudo guix system reconfigure’ and even rebooted, but it seems the offloaded build still fails. 🙁
<apteryx>I had this working before and never needed to play with the option... so something changed or is amiss
<lilyp>apteryx: I'm still positive you could build something with guile's module reflections
<lilyp>also, does refresh work for variables reference via @@?
<davidl>zamfofex: thanks for the npm-import.scm link. Good work there!
<rekado>zamfofex: only the childhurd’s daemon needs to run with --disable-chroot
<f1refly>is someone here using kdeconnect? when I try to start one of the frontends I get this:
<f1refly>error activating kdeconnectd: QDBusError("org.freedesktop.DBus.Error.Spawn.ChildSignaled", "Process org.kde.kdeconnect received signal 6")
<rekado>efraim mbakke I pushed the “update-vulkan-headers-staging” branch a while ago
<rekado>I’m not sure if ci.guix.gnu.org actually built it when I edited the specification from “update-vulkan-headers” to “update-vulkan-headers-staging’
<Luk6655>Hi,What is the recommended workflow for modifying a package source, then installing that modified source? Create own channel and define packages that refer to modified source or some other way?.
<rekado>it would be good if we could prevent this branch from going stale.
<zamfofex>rekado: The childhurd does have ‘--disable-chroot’ that it (indirectly) inherits from the ‘guix-configuration’ in ‘%base-services/hurd’. I am able to build packages on the childhurd, it’s just offloading that fails mysteriously.
<zamfofex>Luk6655: You can also use ‘--with-patch’ if it makes more sense to you.
<Luk6655>zamfofex: yes, it does. It is a nice option, because it doesn't require having to host a copy of the repo elsewhere.
<zamfofex>I spent a long time yesterday trying to figure out why the childhurd’s ‘--disable-chroot’ flag is not being honored when offloading, and I wasn’t able to figure it out. But maybe I was just too sleepy, I guess. (Or maybe I don’t know enough about the Guix daemon. Or both…)
<zamfofex>Here’s a question: How does offloading work? Does it reuse the childhurd’s existing daemon, or does it invoke a new daemon on the childhurd? Surely the latter would be silly, right? But it’d explain what’s going on, I think.
<cbaines>zamfofex, I believe offloading connects to the daemon, rather than starting a daemon then connecting to it.
***reyman is now known as reyman_aw
<Kabouik>What would I need in my config.scm to specify arguments to particular kernel modules? Other owners of my machine say I need to set `snd-intel-dspcfg` with `dsp_driver=1` to get audio working, and also `fbcon=rotate:1 video=DSI-1:panel_orientation=right_side_up` to GRUB config to properly rotate the touch coordinates. I am not comfortable with customizing the kernel, let alone in Guix, and since this fiddles with GRUB, I am afraid of potential
<examors>I am not sure about the GRUB config, though.
<antipode>Kabouik: maybe the 'kernel-arguments' field of operating-system?
<antipode>I'm not aware of a mechanism for passing arguments to _particular_ kernel modules (neither I am saying it doesn't exist)
<zamfofex>A couple updates: I found out that builds were indeed also failing on the childhurd itself! I made sure that ‘--disable-chroot’ is explicited in ‘extra-options’, so it seems to just not be honored for some unknown reason. (???)
***wielaard is now known as mjw
<pkill9>zamfofex: what are you doing with the Hurd?
<zamfofex>pkill9: For now, I’m just trying to get it to build packages.
<Kabouik>Thanks examors and antipode, but that doesn't allow specifying an argument for the kernel module, does it?
<Kabouik>I am really lost with this, I'm not sure I'll be able to fix sound and touch coordinates on that machine.
<examors>Kabouik: A modprobe config file lets you pass module arguments. As shown in the docs, you can create a /etc/modprobe.d/whatever.conf with simple-service, and add `options snd-intel-dspcfg dsp_driver=1` in there.
<zamfofex>A couple more notes: I tried to specify ‘(extra-options '("--not-an-option"))’ to the Hurd VM’s ‘guix-configuration’, and it didn’t try to build a new image for it, so I’m led to believe it is simply being ignored for some reason.
<civodul>zamfofex: weird; according to (gnu services base), 'extra-options' is taken into account
<zamfofex>civodul: Yes, I’ve been looking into it, and I see no reason why it wouldn’t. Though I will say: I noticed changes I apply to ‘guix-configuration’ in general do not cause an image rebuild. But for e.g. ‘openssh-service-type’, they do.
<rekado>those will delete all but the latest generation
<Luk6655>I see, ok, cool, I'll just read about them first, but hopefully they take care of it, thanks :-)
<Kabouik>Unfortunately, either that didn't work as expected, or this fix is not enough to solve the issue on Guix (others reported they just needed this on other distros). I already had sof-firmware installed but still observe the same symptoms.
<Kabouik>Can I check if the module is actually enabled and the configuration options are used?
<Luk6655>rekado: this -d seems to delete all, but the latest generation, I need those 3 system generations (various kernels etc) Is there any other way to discover what's taking so much space?
<Luk6655>those 3 generations really only differ in the kernel
<examors>Kabouik: if it worked I think you should be able to see something in /sys/module/snd_intel_dspfg/parameters/
<examors>Kabouik: You could try setting it via a kernel argument. I don't know if that works for non-builtin modules, but worth a try I guess. Just `(kernel-arguments (cons "snd_intel_dspcfg.dsp_driver=1" %default-kernel-arguments))` I think
<Luk6655>not much... only about 300MB for the user
<Kabouik>Does that go under (operating-system) examors?
<Kabouik>Awesome. That Friday is ending well, finally.
<Luk6655>at the moment I'm just running mate-disk-space-analyzer on /gnu/store to see what's taking so much space there. BTW, what are the sizes of /gnu/store of people that have been running guix for a long time (months)?
<podiki[m]>for any matrix users: the libera chat bridge still seems to be one way here, you can send but won't receive (non-matrix) messages; I believe you'll see this message since I'm connected via matrix
<examors>Kabouik: I think you can actually add those other parameters to kernel-arguments too: `(kernel-arguments (cons* "snd_intel_dspcfg.dsp_driver=1" "fbcon=rotate:1" "video=DSI-1:panel_orientation=right_side_up" %default-kernel-arguments))`
<lilyp>in that case, you might be stuck in the moderation queue (to which I don't have access)
<dthompson>style question: when a gexp in a package refers to other packages, should those referenced packages be listed again in the package inputs? I know that technically speaking they don't need to (unless of course there is build code that requires that package be in the inputs list)
<Luk6655>zamfofex: I thought about it, I'll run du on them first.
<zamfofex>I think you should run ‘guix size ~/...’ instead, that might make more sense.
<Luk6655>yes, probably, do comes back with all zeros
<Luk6655>guix size doesn't work (or I'm using it wrong): guix size: error: /home/luk/.cache/guix/profiles/ltxapehp5jvjsxmkareeyd2lbkrtuo2vtukrtknkktacbk5fwyrq: unknown package
<zamfofex>I guess you need to use the store path they link to.
<Luk6655>each of the folders has a manifest file in there that specifies the packages that were used to create it, so this allows me to identify them, the file also includes the store location hopefully this will work with guix size
<zamfofex>This works for me: ‘guix size $(realpath ~/.cache/guix/profiles/*/)’
<Luk6655>so now the question becomed, do I just delete the folders and does this mean everything will get garbage collected? or will those items stay "in limbo"?
<Luk6655>zamfofex: yes, it works for me to, they come to about 10GB all together, so not that much considering the store is 45gb
<Luk6655>on the subject of clearing them up... I might just try to delete one and see if gc clears what was in it
<Luk6655>well, it doesn't appear to delete stuff... or perhaps there are other profiles that refer to the same packages that were specified in the profile cache I deleted
<Luk6655>Does anyone know for sure If I delete those cached profiles it will be able to garbage collect those packages from the store?
<Luk6655>or is there "a proper way" to remove them?
<Luk6655>there appear to be only links in the cache folder to /gnu/store/xxxxxxxxxxxxxx-profile locations, if I delete the links this seems to leave the profiles untouched
***mark__ is now known as mjw
<Luk6655>I can still run guix size on them despite the deletion of the link, and gc frees nothing more...
<VesselWave>Hello, I've submitted a patch to guix https://issues.guix.gnu.org/issue/57702 . As I use protonmail I couldn't sent patch through 'git email-send' for free. I've sent manually through web interface. As turned out I should have used plain text option in protonmail. So what could I do: create new issue correctly with plain text or maybe I can fix this one?
<rekado>VesselWave: just send a new one to the same address.
<rekado>VesselWave: also note that we prefer to have one commit per change.
<Luk6655>I thought that perhaps guix package --list-profiles can show those profiles and be able to delete them somehow, nope it only shows two profiles. If no one knows how to delete them perhaps this is a bug in guix?
<rekado>VesselWave: you don’t have to, but if you want you can just send a reply to the issue email address, and then send your patch after that.
<Luk6655>guix gc -d, no, I tried guix pull -d and guix package -d (as normal user, as I didn't want to delete my 3 system profiles), then I run guix gc, both as user and root
<jpoiret>your system profiles can be deleted with `guix system delete-generations` only
<rekado>Luk6655: if your root user has some profiles you should delete their old generations too.
<jpoiret>also, guix gc doesn't need to be run as root
<Luk6655>rekado: root only has two profiles, its user profile that contains vim only, and the system profile. Root's account is only ever used to run guix system reconfigure and to edit an odd file with vim.
<klm_>apteryx: Thanks, but I haven't quite been able to grok the desktop service definitions. I don't want to run a desktop manager, and there seems to be little documentation about this. xinit / startx don't work.
<klm_>Fot reference, I added (extra-special-file "/usr/bin/my-start-x" (xorg-start-command ...)) into my services which worked