<nckx>Thanks for the tip regardless, I really should try it, even though Rust.
<apteryx>seems libusb is not working as intended on Guix System? GNOME Boxes for example isn't able to access USB drives (libusb: error [get_usbfs_fd] libusb couldn't open USB device /dev/bus/usb/001/089, errno=13)
<dstolfa>mbakke: i generally don't like installing these replacement utilities because when i ssh into places that don't have them, i hate myself for doing it because i forget how to use the base utilities :D
<nckx>I like raghavgururajan but do I want to think of them every single time I grep for a thing is the thing.
<nckx>Here I am grepping tiny appz like ‘Linux’, not full operating systems like Chromium.
<mbakke>I was actually surprised how fast rg entered my "muscle memory", yet I still use 'grep' for binary files (because I can't be bothered to learn how to include those with rg), as well as pipes (because grep is always available and probably faster in that case)
<drgibbon>It is on my VM install, maybe I made a mistake somewhere though.
<emestee>iirc during the boot it only needs to ask you for it once
<drgibbon>ahh ok, well I'll see how I go when I install it on bare metal.
<raghavgururajan> /home/rg/.guix-profile/bin/chromium: error while loading shared libraries: /gnu/store/6yn16h7la1cj64gm5gvklhijcd4k1zgb-cups-2.3.3/lib/libcups.so.2: file too short
<raghavgururajan>nckx: ^ Also because of store corruption? The above error persists even after `guix gc --verify=contents,repair`.
<tissevert>drgibbon: at what steps do you have to enter it twice ? grub + normal mount during early init ?
<drgibbon>tissevert: First before seeing the GRUB menu, and then second after selecting a GRUB option.
<tissevert>yeah, I think I had the same when I tried to install with my whole system encrypted
<tissevert>sounds a bit inevitable: grub has to unencrypt the partition to find the kernel and boot it, but then I don't see a lot of options to pass the key securely to the kernel : procline is out of the question, write it to a file on /boot ? what if it's read-only ?
<drgibbon>Is it possible to boot off of an LVM volume? So one big block that GRUB unlocks, then finds the /boot inside an LVM setup?
<tissevert>I have no idea how LVM works and if GRUB knows how to handle this
<jonsger>hm, libaio fails on core-updates. thats only some dependencies until I can built firefox 89 :(
<drgibbon>Definitely possible to have a bunch of LVM "partitions" inside one big encrypted block (LUKS/LVM), but I'm not sure if GRUB can handle that. I've only ever done that with a separate unencrypted /boot partition.
<tissevert>raghavgururajan: that's great ! thanks for the confirmation
<munksgaard>How do I best combine guix time-machine and guix environment? I want to set up a reproducible environment for a project of mine, and I have used a manifest file in conjunction with `gnu environment -m ...` so far, but now I also want to make sure that the channels are the same. It doesn't seem like the output from `guix describe -f channels` go in my manifest file though?
<munksgaard>I'm guessing I have to use `guix time-machine` instead, but it's not clear how I can provide a manifest file as input...
<rekado>munksgaard: you can skip the “time-machine” command and use inferiors instead, right from the manifest
<rekado>or you could do “guix time-machine -C channels.scm -- environment -m manifest.scm”
<rekado>i.e. load the channels from channels.scm to tell “guix time-machine” what variant of Guix to use, and then use the manifest with that version of Guix.
<munksgaard>Aha! That's very helpful indeed. Thank you rekado
<rekado>which method you pick depends on whether you want the channels to be distributed separately and independently of the manifest.
<rekado>personally, I like to keep them together in the same file
<rekado>note that using inferiors is significantly slower than just running a manifest through the “current” Guix (whether that is ~/.config/guix/current/bin/guix or a Guix variant spawned by “guix time-machine”)
<jonsger>mbakke: I have a change for libaio on core-updates. Can I directly push or should I go over the mailing list?
<dstolfa>i'm trying to start a service, and the only way it could be started is if i have STRONGSWAN_CONF=... set to the new derivation, but even with guix utils in use-modules it's complaining about with-environment-variables being unbound, which i assume is due to me trying to call it in a gexp
<nckx>with-environment-variables is merely syntactic hac^Wsugar to make environment variables appear ‘functional’, but you can also just (setenv "FOO" "bar") if you don't care about restoring the original environment, and it sounds like you wouldn't.
<dstolfa>ah i see, so just (setenv "..." "...") would work for all the following (invoke ...) calls?
<dstolfa>nckx: i have one bug leftover, and i figure it's easier to just ask. when i do an `ipsec stop` from the command line, the whole daemon gets killed (including all the subdaemons like charon). however, when i do (invoke #$ipsec "stop") the other daemons stay running.
<dstolfa>is there a way to pop a guile repl or some kind of debugger on `herd start/stop` for a particular service? for some reason a command that runs successfully doesn't do what i would expect it to do, and i don't know how to debug it
<roptat>well, matrix was the way for people to ask questions and presenters to coordinate, but it was all managed by a jitsi instance for presenters, which was then streamed live, with a widget in matrix to be able to watch the stream
<roptat>we might be able to do something similar if we can set up a stream, and use IRC or matrix as the communication channel
<katco>note that matrix has native integration for starting jitsi conferences in matrix rooms
<katco>that was worded poorly. i should say that the reference client, element, has native integration for starting jitsi conferences from a matrix room. i know nothing about what support exists at the protocol level (i.e. matrix)
<avalenn>katco: there is none at the protocol level for group conferencing
<avalenn>Matrix can serve as signaling protocol for one-to-one WebRTC discussions, but no more
<avalenn>And it is not a priority of Matrix development at the moment as the client-side integration with Jitsi (and I wonder if there is one with BBB too) seems to be sufficiently useful for Matrix users
<tissevert>and thought instead I'd ask about the promising documentation strings
<tissevert>because then I'd probably be able to fish more efficiently myself
<dstolfa>sneek: later tell nckx: so i've got strongswan starting correctly, setting up the (legacy) configuration and all that, but when i run `herd stop strongswan`, i see that `$STORE/sbin/ipsec stop` gets run successfully (it returns #t), but it doesn't actually stop the daemon. if i run it from the shell, it works as expected. the stop gexp is literally (invoke #$ipsec "stop"), am i missing something?
<sneek>nckx, dstolfa says: so i've got strongswan starting correctly, setting up the (legacy) configuration and all that, but when i run `herd stop strongswan`, i see that `$STORE/sbin/ipsec stop` gets run successfully (it returns #t), but it doesn't actually stop the daemon. if i run it from the shell, it works as expected. the stop gexp is literally (invoke #$ipsec "stop"), am i missing something?
<nckx>raghavgururajan: Corruption runs rampant-ish in your store! Did you add ,repair?
<nckx>dstolfa: Do you know how ‘ipsec stop’ actually stops the daemon? Does it read the PID from somewhere? Prod a socket?
<rekado>this is not difficult. It’s just writing to and reading from a file. But every time I look at this my movements slow, my eyes feel like I steeped them in whiskey for hours, and any attempt to speak just produces grunts.
<rekado>nckx: I think the containerisationisation was done because we could.