IRC channel logs


back to list of logs

<maximed>It appears in open-luks-device in (gnu system mapped-devices) (possibly other places as well)
<maximed>ass #$(file-append cryptsetup-static "/sbin/cryptsetup")
<maximed>* as
<maximed>so the full name "/gnu/store/.../sbin/cryptsetup" ends up hard-coded in some initialisation script
<maximed>Fare: FWIW, the ‘guided graphical installer’ supports cryptodisks IIRC
<maximed>dunno about LVM though
<jackhill>Fare: maybe also post your config.scm?
<Fare>maximed, I want not to erase the rest of the data on disk, so I'm wary for installers.
<jackhill>what's interesting to me is that it seems the configuration is correct enough that grub can find the kernel and initramfs from the store…
<Fare>I'm experimenting some more. jackhill, where in your config do you tell the initrd to include cryptsetup ?
<jackhill>Fare: I don't explicity, as I understand it, it's supposed to figure it out becasue the mount-point for / has a dependency on mapped devices. I do see a bunch of warnings about /dev/sda2 might not be mounted, but it seems to work the two times I've tried it. I'd easily believe we have bug where it only works sometimes due to some factor we haven't taken into account.
<jackhill>Fare: if you have a spare machine/vm you could try it with just luks and no lvm, to see if you could get that part working perhaps?
<Rooks>how do adjust a package in order to use a specific source for building a dependency?
<drakonis>you have to switch the dependency?
<jackhill>Rooks: look at package transofrmation options: In what context do you want to do it? For example, with `guix build --with-source=dependency=url foo`
<Rooks>Trying to write a package file, and installing it to a profile
<Rooks>Well, currently trying the default profile
<jackhill>ah, there's difining varient packages: The lua example there might work for you? Or look in guix for calls to package-inputs for manipulation of inputs
<jackhill>have to run, sorry
<raghavgururajan>Fare: I am currently using LUKS+LVM setup. Here is my config
<Fare>I feel like I'm rebuilding things that should already been built... how do I tell whether caching is working properly or not?
<maximed>Fare: what do you mean exactly?
<maximed>If you're referring to substitutes: "guix build minetest" should get you a substitute of Minetest, without having to build anything locally
<maximed>(if substitutes are enabled)
<the_tubular>Where can I read more about guix VM ?
<apteryx>info '(guix) Invoking guix system'
<apteryx>then C-s (Control + S) to search for vm
<apteryx>I was pondering about turning the 'target' field of the bootloader-configuration into 'targets', to accept a list of devices to install the bootloader on (/dev/sda, /dev/sdb), but reading the manual I see that for UEFI systems target is a path to a mounted partition such as '/boot/efi'; so it seems it is device agnostic in this case (RAID for example would be taken care at a lower level, so you
<apteryx>wouldn't need to pass multiple targets, IIUC).
<apteryx>uh, seems you can't have /boot/efi on soft raid1:
<apteryx>so I guess targetS for UEFI would be useful after all. You could have /boot/efi from /dev/sda mounted as the 'main' one, and /boot/efi-backup mounted as a fall-back from /dev/sdb. You'd pass these two mount points as targets to have them in sync.
***bear_ is now known as mallarme
<bost>I'd like to make a few manual improvements. Can you tell me please where's is the source code and/or how to make a pull request?
<bost>I mean documentation improvements.
<leoprikler>patches are sent to but don't forget to read `info "(guix)Submitting Patches"' before that
<bost>JFR the documentation is here
<arjan>does anyone have recent experience with getting guix system to run on an armhf board like beaglebone black?
<arjan>i am trying to crossbuild the installation image from source right now using qemu-binfmt but it is taking a long time and tests are failing for quite some packages
<arjan>so not sure if i am on the right track
<bricewge>arjan: I tried month ago to build Guix System for the BeagleBone Black, but without success
<rekado_>recently I got a Rock Pi S with 512MB RAM. I’m not using Guix System but Guix-built software on top of it.
<bricewge>I would guess you are on the right track
<rekado_>(I’m using aarch64 packages, not armhf)
<arjan>for now i am just disabling tests for the packages where they fail to keep the build going
<arjan>aarch64 seems to have more substitutes available than armhf
<arjan>bricewge: did you crosscompile and was it the installation image or another system configuration?
<bricewge>I used the operating-system template from the repo and complied it through qemu-binfmt
<arjan>ah do you remember where it failed?
<bricewge>rekado_: That's a cute SBC
<bricewge>arjan: No sorry
<arjan>alright thank you!
<rekado_>bricewge: it is! I’m using it as a piano MIDI instrument (with liquidsfz).
<pkill9>bricewge, rekado_: what are you talking about? is there a pic of something?
<rekado_>pkill9: we’re talking about the Rock Pi S.
<rekado_>it’s a very small aarch64 board.
<bricewge>I have the following error when using "guix deploy" « error: failed to deploy host: remote command 'guix repl -t machine' failed with status 25 »
<bricewge>And I have the following on the depolyed machine log « pam_elogind(sshd:session): Failed to connect to system bus: No such file or directory »
<bricewge>Any idea what's wrong here?
<dstolfa>is the guix manual down?
<dstolfa>hmm... i wonder why my DNS doesn't want to go to
<oriansj>dstolfa: has address
<dstolfa>oriansj: thanks for confirming. i think my router's DNS fell over, a simple restart fixed it
*dstolfa shrugs
<oriansj>also always have local copies of manuals and source-code
<Noisytoot> has no IPv
<rekado_>data centre switch limitation
***lukedashjr is now known as luke-jr
<zacchae[m]>So in the installation guide, it has you `herd start cow-store /mnt`. Is this important when writing an image from an existing install?
<dstolfa>zacchae[m]: i'm tempted to say yes
<zacchae[m]>It seems important, but when I try, I get `herd: service 'cow-store' could not be found`.
<zacchae[m]>and `cow-store` doesn't show up anywhere else in the user manual, or in a `guix search`
<zacchae[m]>whelp, I'm trying it. It's going super slow, but Imma just cross my fingers
<cage>Hi, I have installed, on guix-sd, gcc using 'guix installgcc-toolchain' but seems i get no alias for 'gcc' as 'cc', is this intended or i am doing something wrong?
<cage>$ cc -> -bash: cc: command not found
<zacchae[m]>did you source your guix profile after installing?
<zacchae[m]>easiest way is to restart your terminal/tty
<cage>i think i update my profile (i put the suggested lines on my .bashrc)
<cage>ok i'll try to logout/login as you suggested, thanks!
<zacchae[m]>oh wait, isn't cc a different compiler for gcc? if you want `gcc` to run when you call `cc`, jush add `alias gcc=cc` to your .bashrc
<cage>ahh i see!
<cage>ok i am trying to write a definition package for a software that assumes that alias, i guess i need to set the alias in the file there too
<cage>zacchae[m]: thanks!
<zacchae[m]>ok. I don't think scripts respect aliases by default, but I think it is possible. good luck
<cage>zacchae[m]: seems that just setting (setenv "CC" ,(cc-for-target)) in the 'configure phase does the job :)
<cage>now i would like to submit the patch, the software is a gemini server and now i wonder if is the web.scm is the right file where to put the definition
<cage>the author of the software would prefere net.scm and i agree
***Kimapr4 is now known as Kimapr
<bricewge>zacchae: Where did you run that herd command, on the installer image?
<bricewge>It's defined in `(gnu system install)` but not exported as just a service
<zacchae[m]>no, on my normal install. I just wanted a normal system install on a flash drive
<zacchae[m]>So I need that to write an install to another drive?
<zacchae[m]>I'm assuming I don't need the installer image for this, because that would be strange
<bricewge>You are trying to build a custom installer image?
<zacchae[m]>I mean, I would be interested in that, but currently I am trying to make a normal install on an external drive
<bricewge>Then you don need to bother with this service AFAIU
<zacchae[m]>I hope I'm doing it right. It is being REALLY slow, but hasn't crashed
<cehteh>how do i change the X-cursor size in guix?
<bricewge>zacchae: Did you followed ?
<bricewge>Specially the image types iso9660 and efi-raw
<bricewge>cehteh: It depend on your window manager, it's done as in standard linux distro
<bricewge>By setting it in `.Xresources` for example
<cehteh>ok i look into that later
<raghavgururajan>Hello Guix!
<zacchae[m]>bricewge: hmm. I'm curious what the difference between installing on a drive vs writing a volitile image to a drive is, but it looks like the image is what I want
<zacchae[m]>hello raghavgururajan!
***luke-jr- is now known as luke-jr
<abrenon>I don't understand the repo option in (guix import utils)' recursive-import
<abrenon>it looks like the repo option plays a special part
<abrenon>and I wonder why
<abrenon>why is this option special, why can't any option whatever its name gets passed recursively
<iskarian>abrenon, do you mean the #:repo argument in the #:repo->guix-package proc?
<iskarian>Oh, I see it's actually passed in to the proc by recursive-import
<iskarian>It's so it's easier to use the same code to import from compatible repositories
<abrenon>yeah, the very same
<abrenon>I don't understand why this key would be so important that it deserves a special handling
<iskarian>(for example if someone hosted a mirror, or a "free software only" version, etc.)
<abrenon>in particular, I was planning to change this very key for opam, to allow passing a list of repositories
<abrenon>and I can't because of this key, and I really can't figure why all importers would share the common feature that they need to expect one repo
<abrenon>it's a bit weird if you ask me : )
<iskarian>Oh, repo is passed as-is from recursive-import to the proc, so I suppose you could put a list there, or anything you want
<abrenon>yeah, but I don't like to call repo (singular) a thing that holds a list of them
<abrenon>and I wonder why it needs to call the poor importer with that option set, even if none was passed
<abrenon>and, on the other hand, not passing any other options
<abrenon>like, why can't custom options be passed to the importer in recursive mode ? ^^
<iskarian>typically that's done by passing a repo->guix-package which closes over them
<abrenon>that's clever, but then why explicitly pass this repo option at all cost like it was something necessary, that still is very unexpected I think
<abrenon>well, it works and that's the most important, I just see that I'm puzzled again by things I find unexpected
<iskarian>You'd probably have to ask "wigust", who added recursive-import
<abrenon>thanks for your explanations though !
<iskarian>I really do think that the intention was for repositories to be pluggable on a high-level, even if that never made it into the actual CLI
<abrenon>pluggable ? how ?
<iskarian>For example, say someone forked a version of CPAN, which contained a different set of software; then you could import from the fork by just changing that repo argument
<iskarian>(which could in theory be supported on the command line, say with --repository=)
<abrenon>oh, I think I see, you mean it could have become an option of the guix script "import" itself, not of any particular one like "opam" or "cpan" ?
<iskarian>Actually, #:repo *is* used, by the CRAN importer, with the --archive= argument
<abrenon>: )
<abrenon>well, anyway my importer works so that's a start
<iskarian>ah, a new importer? What is it for?
<abrenon>now I'll have to have the imported packages to compile… but that's another story
<abrenon>no, no, sorry, it's not new
<abrenon>I added support for custom repositories in opam's
<abrenon>and for several sources at once, so that it doesn't forget about the main opam simply because it's importing from, say, coq
<abrenon>thanks again for your insights, and see you !!
<iskarian>ah, still good to see work on the importers! It's just too bad they can't autogenerate good descriptions...
<iskarian>have a good one :)
<dstolfa>this might be a stupid question, but is there an ETA on the core-updates merge?
<maximed>dstolfa: It (core-updates-frozen) will be merged once most build failures are fixed I presume
<dstolfa>ah okay, cool. thanks!
<maximed>ETA: it will be earlier if you help. I can't give a concrete date
<dstolfa>maximed: there are many guix things i want to do, i'm just really pressed for time right now :-/. i've been working 14-16 hour days
<retropikzel>I'im installing guix and I need to install grub with --removable flag, what should I add to (bootloader-configuration...?
***namtsui_ is now known as namtsui
<bricewge>Looks like cuirass can't take a channel record
<bricewge>That's odd
<bricewge>`channel->code` to the rescue!
<rovanion>How come gcc does not provide a cc binary?
<rekado_>rovanion: upstream does not. We don’t generally install binaries upstream doesn’t install, nor do we rename them.
<rovanion>Any tip on how to add an alias/symlink in a Guix Environment?
<ssouth>rovanion: Not certain but I believe this is what the "special-files-service-type" is for.
<ssouth>Or perhaps "extra-special-file" would be more appropriate.
<rovanion>ssouth: I'm reading that page to be about services as part of an operating system rather than an environment. Using `guix environment --load=test.scm" where test has the following content does not work:
<ssouth>rovanion: Ah, I assumed you were working with a system configuration.
<ssouth>What you have certainly will not work. But are you not able to create a symlink directly in the filesystem?
<ssouth>Using 'ln -s', etc.
<ssouth>rovanion: Are you simply trying to have "cc" recognized in place of "gcc"? Perhaps "alias cc=gcc" is all you need.
<rovanion>ssouth: Yeah that will do, will just have to figure out how to put that in .envrc-files.