IRC channel logs


back to list of logs

<nextos>hi, im installing guixsd for the first time on bare metal. It seems to be taking a bit of time as lots of things (like the nix store) are getting compiled. Why is this not distributed as binary?
<sneek>nextos, you have 1 message.
<sneek>nextos, rekado_ says: You can use Guix as a package manager on top of any other distro (including Arch).
<ZombieChicken>nextos: GuixSD can handle both. The install docs do (or should) give you instructions on how to enable binary downloads from hydra
<nextos>ZombieChicken: aha, i missed that, i come from nix and i thought it was enabled by default, thanks
<cbaines>Evening all, does guix system reconfigure need to be run as root? I'm using guix from the git repository as a non root user, and I think if I just use sudo, it doesn't use ~/.config/guix/latest
<cbaines>If I try running just as my normal user, I get a permissions problem with creating a new profile
<Apteryx>Hi! I'm trying to get Japanese characters displayed in IceCat, so i've followed the manual and installed the font-adobe-source-han-sans:jp; helas it still doesn't work (even after a reboot). Any idea?
<buenouanq>guix package -i $(guix package -s ^font- | grep name: | sed s/name://g)
<buenouanq>would be great if I could just do something like guix package -i font-* or something
<buenouanq>is there a reason it doesn't offer wildcard expansion like that?
<buenouanq>anyway Apteryx, I don't know which package it was, but installing all the fonts works
<buenouanq>still haven't been able to get ibus-anthy to work for me though
<Apteryx>buenouanq: This will pull in all the fonts, right?
<nextos>Another question, i didn't set up a dhcp service on my first configuration. I have then rebooted my machine, and connected to Internet by manually calling dhclient. However, calls to guix reconfigure fail with "In procedure getaddrinfo: Name or service unknown". How can i get guix to use this temporary connection?
<buenouanq>well, any package that begins with the string `font-', yes
<buenouanq>nextos: you have to do the whole \\ $ ifconfig <interface> up \\ $ dhclient -v <interface>
<buenouanq>you ping or something to make sure it's actually up?
<nextos>its up
<nextos>buenouanq: its up
<nextos>buenouanq: oh, now its connecting. Seems hydra was unreachable for a while
<nextos>from my place. Crappy connection...
<Apteryx>buenouanq: Well, you could use regexp to do the `guix package -i "regexp"`, I believe.
<Apteryx>maybe "font-.*" ?
***quasisan1 is now known as quasisane
<ZombieChicken>sneek: help
<erliphant>does anybody work with guix in an environment with nfs home directories? how do you deal with the fact that /gnu/store may not be available?
***atw` is now known as atw
<baconicsynergy>I can't find a substitute for Kodi, but when I attempt to build it from source, it says it can't find nasm, even though I have it installed :/
<baconicsynergy>anybody else encounter the issue? it cuts off riiiight when its spits out "checking for"
<rekado>sneek: later tell baconicsynergy When building things with Guix it doesn’t matter what you have installed locally. All that matters is what the package expression declares as inputs, as these are the only things that appear in the build environment.
<sneek>Got it.
<sneek>Welcome back Helius, you have 1 message.
<sneek>Helius, rekado_ says: rhdf5 is a bit anomalous. I would patch the libhdf5ForBioC.a target in src/Makevars to not unpack the tarball and unpack it in an extra build phase (after “unpack”). It’s pretty unusual to have modified upstream sources bundled as tarballs. I’m surprised this passed bioconductor peer review.
<Sleep_Walker> - are there some new dependencies required for build on top of other distro?
<rekado>Sleep_Walker: the build is not failing, is it?
<rekado>Sleep_Walker: there are optional dependencies.
<Sleep_Walker>well, the build returns no-zero value, I though this is the reason
<Sleep_Walker>but yeah, you're right
<Sleep_Walker>it's getting to make install phase
<Sleep_Walker>damn, I'm slow today
<stavros__>is it possible to run GuixSD on Vbox?
<bill-auger>probably - there is a qemu image you could convert
<jmd>stavros__: I don't know. Why don't you try?
<bill-auger>but if you want a GUI for qemu you could try virt-manager
<stavros__>I think it is possible to load a qemu image in vbox
<ng0>How reliable is our crate importer? With my very limited knowledge of rust: an import I just made references input rust-libc. Is this rustc by any chance, or is there really a package named libc in crates?
<ng0>importing libc says:
<ng0>A library for types and bindings to native C functions often found in libc or
<ng0>other common platform libraries.
<ng0>okay, got it
<ng0>welp.... so much for "just" packaging the prototype
<ng0>I estimate I will throw a giant dependency graph to the guix-devel list in 1 or 2 months
<ng0>The joy of being the first one in a language to package
<jmd>ACTION manages to install GuixSD without ever touching a shell.
<ng0> for bugs? I can't search at the moment
<ng0>i wonder why there is winapi in the depndency graph on a lower level...
<ng0>95 rust packages upcoming soon'ish
<Helius>rekado: thank for the email
<Helius>It’s Raoul
<ng0>are any people working on rust packaging? If so, feel free to message me offlist so I can share what I need to package in the next months so we can avoid duplicate work
<jmd>I'm not.
<jmd>ng0: A short message to might catch more people ...
<ng0>I'll do it later
<ng0>you're right
<ng0>will we follow the python path and throw libraries into rust.scm ?
<ng0>I will do it this way
<ng0>complains can be filed later at review
<ng0>so the crate importer puts "crate-uri" into the package definition, but crate-uri is unbound
<ng0>is this because we haven't actually packaged/included cargo? I updated my branch but it's not there yet
<ng0>oh, got to import the actual build system
<cbaines>Is there any prior art in running guix system reconfigure not as root? I can build the system using guix system build, but for various reasons, I'll get a system if I run the build as a different user
<cbaines>If there is some way to reconfigure to a system directory in the store, that would work well?
<suitsmeveryfine>Hi there! I try to install guix SD with encrypted /home but fail every time. The problem is that, for some reason, /home isn't mounted after boot even though I enter the LUKS password
<sneek>suitsmeveryfine, you have 1 message.
<sneek>suitsmeveryfine, rekado says: This is not a bug. When you use the git checkout and you run configure inside of ‘guix envirnoment guix’ a reference to the current guile executable will be embedded. There’s no gc root for it, so it will be GC’d. To fix this you only need to run the configure script again.
<suitsmeveryfine>so the login screen appears but I'm unable to load e.g. Xfce
<cbaines>suitsmeveryfine, do you see any messages?
<suitsmeveryfine>no, not really. If I change the TTY and log in as root I don't find my user under /home
<suitsmeveryfine>so something goes wrong there
<suitsmeveryfine>The thing is, I'm using a configure file that worked fine when I last installed this distro maybe six months ago
<suitsmeveryfine>I guess something has changed within Guix SD since
<cbaines>Hmm, very possibly
<cbaines>have you found a workaround, maybe running "mount -a" would mount /home?
<jmd>Whilst trying to install GuixSD, when I attempt to start cow-store I get:
<jmd>herd: exception caught while executing 'start' on service 'cow-store':
<jmd>ERROR: In procedure rmdir: Device or resource busy
<suitsmeveryfine>cbaines: no, this doesn't work; also, I fear that the installation wasn't completed properly since
<suitsmeveryfine>the symlinks must be created inside the user's directory, right?
<cbaines>jmd, so rmdir is called on /.rw-store as the last step of starting cow-store
<cbaines>I don't quite understand the service yet, but could it be that the service is somehow already running?
<jmd>cbaines: Possibly. Although I thought that properly designed services should prevent that from ever happening.
<cbaines>jmd, well, you may have encountered a bug in the cow-store service
<cbaines>its a service that I think is just used during installation
<cbaines>have you tried stopping it, and the starting it?
<jmd>It is not running (according to "herd status")
<cbaines>suitsmeveryfine, unfortunately, I'm not familiar with setting up Guix with an encrypted /home
<cbaines>I did manage to get Guix installed yesterday will full drive encryption, which I'm pretty happy about :D
<suitsmeveryfine>cbaines: oh, I wouldn't mind doing that
<suitsmeveryfine>Will I need to create a /boot partition for that?
<cbaines>jmd, if your happy trying to fix this, I'd recommend checking if /.rw-store exists, and then trying to remove it manually before starting the service
<suitsmeveryfine>cbanes: would you mind sharing your config file for encrypted root?
<cbaines>suitsmeveryfine, I didn't, although I do have to enter my passphrase twice, once for grub, and then once again
<suitsmeveryfine>cbaines: ok, I don't run libreboot on this particular machine
<jmd>cbaines: Yeah. The problem is that /.rw-store seems to be "busy"
<suitsmeveryfine>so I guess that I can't encrypt everything
<cbaines>jmd, It might be a mountpoint for something, and running mount might tell you what
<cbaines>jmd, unmounting it may help
<jmd>It's a mount point for unionfs
<cbaines>suitsmeveryfine, why do you mention libreboot? I don't think that is required for full drive encryption
<suitsmeveryfine>cbaines: how else do you reach the Guix SD Grub screen?
<suitsmeveryfine>or do you put it on a USB drive?
<cbaines>suitsmeveryfine, I'm just using the BIOS/UEFI (I don't know which applies to this machine) to get to grub
<cbaines>As for the config file, rather conviniently, I just had to make a few small modifications to the template
<cbaines>The mapped-devices part of the operating-system record is what is the most relevant bit for full drive encryption
<suitsmeveryfine>cbaines: What I mean is that if the entire HDD is encrypted then how can you reach the point where you enter the LUKS password?
<cbaines>jmd, does unmounting /.rw-store and then trying to start the service help?
<cbaines>suitsmeveryfine, ah, its just the parition that is encrypted, so I have 3 partitions, where 1 of those is the root partition for GuixSD, and that is encrypted
<cbaines>I ended up following the Gentoo wiki for the partitioning guide, and in doing so I created a BIOS boot and EFI system partition as well
<ng0>so I'll end up hacking on cargo-build-system before I can get it all going
<ng0>first package I encountered the "no cargo.toml" problem
<jmd>cbaines: It seems to help, yes.
<suitsmeveryfine>cbaines: so the template config file is in fact unusable as-is
<cbaines>suitsmeveryfine, Yep, its just a template, the grub configuration needs changing, and the mapped device configuration needs changing (to add the UUID and set the target), and then the file-system configuration might need changing also
<jmd>cbaines: But that seems odd, because in the cow-store-service, the thing immediately before rmdir /.rw-store is mount /.rw-store
<cbaines>jmd, I agree, I don't quite understand what is going on in that service yet
<ng0>if we leave out --bin for when no Cargo.toml is found, what would happen? Or was the idea to just copy it plain into the store?
<ng0>I'm reading through the threads to find more info
<Apteryx>Is lsh supposed to be able to connect to an OpenSSH server?
<cbaines>Apteryx, yep, although you might find it a bit tricky
<Apteryx>cbaines: OK. Yes I know not much about lsh and was just trying to connect to an OpenSSH server in the same way I'd usually do it with (open)ssh, but I get a "lsh: Protocol error: Algorithm negotiation failed.".
<cbaines>Apteryx, sounds famililar, I've encountered that before
<cbaines>I can't remember what I have done, but there is a recommendation to use -c aes256-ctr here
<Apteryx>I just installed openssh to make sure nothing has broken since I last connected, and I can still connect fine using ssh.
<Apteryx>cbaines: Thanks! Got it working :)
<Apteryx>Noticed it asked me for a password too, which was unexpected (I'm using a key). Will double check the sshd_config file.
<Helius>question, which is the simples way to generate a pacth from emacs ?
<jmd>Helius: try #emacs
<jmd>How is the staging build going? (just out of curiosity?)
<nextos>hi, i'm getting a 410, "Gone" error when trying to install texlive-minimal (guix package -i texlive-minimal). Any clue why this is happening?
<nextos>Seems the substitute is not on anymore
<nextos>guix substitute: error: download from '' failed: 410, "Gone"
<lfam>sneek: later tell jmd: Something came up and I wasn't able to start the staging build when I had planned. But it is running now
<sneek>lfam, you have 1 message.
<sneek>lfam, ZombieChicken says: Apparently the virus was a false positive. ZC rechecked the supposedly infected files and they no longer listed as viruses, but the checksums were the same.
<sneek>Will do.
<lfam>nextos: We had to stop serving substitutes of texlive-texmf because it is a humongous package (~4 GB) and we currently lack the resources to compress that and send it out to every user
<lfam>nextos: Use --fallback to build it yourself. This isn't a great loss for individual users, in my opinion, because the package doesn't really "build" anything, it just copies a large amount of data around.
<lfam>The mechanism we use now (returning 410) is not great... we know. We'll revisit when our new build farm front-end is online.
<lfam>I think that institutional users will want to serve their own substitutes and can decide how to handle this case
<rekado>sneek: later tell Helius re creating a patch from Emacs: use git for formatting patches from commits. There’s an excellent git interface for Emacs called magit.
<sneek>Will do.
<rekado>lfam: it’s a bug that guix thinks that this package is substitutable.
<lfam>Speaking of institutional users :)
<rekado>lfam: it is marked as not substitutable, so it’s odd that guix still tries to download a substitute.
<rekado>ACTION –> zzzZ
<lfam>I remember you mentioned this recently, but I don't see the bug report :)