IRC channel logs

<vagrantc>axd-v: right, had the same experience ... i'll boot up my system and see how i've configured it

<axd-v>vagrantc: thank you so much!

<axd-v>vagrantc: I'm gonna be afk for the next 30-40 minutes, but please feel free to post your solution to my problem here. I'm gonna make use of it the second I'm back.

<civodul>rekado: you need a "cvs add" for the last post

<civodul>lemme see

<vagrantcish>axd-v: all i've got is (service libvirt-service-type (libvirt-configuration (unix-sock-group "libvirt"))) and adding the user to the libvirt group

<axd-v>vagrantcish: interesting. Someone earlier today was telling me about some bug and that they needed to have their user as part of the wheel group. I took it to mean that wheel is the only important group and replaced my unix-sock-group with "wheel" and this is the config that isn't working. I'll try adding the user to libvirt and then running the normal config like yours.

<vagrantcish>axd-v: also using %desktop-services

<vagrantcish>axd-v: i haven't tried it with a user not in the wheel group, perhaps they are both necessary? that would seem odd, though.

<civodul>rekado, roptat: should be fixed now (the time for gnu.org to update the page)

<axd-v>vagrantcish: maybe the case

<axd-v>vagrantcish: Changing the unix sock group back to default fixed it.

<vagrantc>ls -l /var/run/libvirt/*sock*

<axd-v>thank you so much for debugging my problems with me. Wheel is no virt.

<axd-v>vagrantc: that returns with libvirt-admin-sock

<axd-v>ah

<axd-v>owned by libvirt group, got ya.

<vagrantc>ACTION wonders why /var/run vs. /run ...

<pkill9>out of interest, can you make a package without any source?

<pkill9>it does't let you omit the source field, or set it to #f, or set the origin to #f, I'm curious to know if it's possible

<pkill9>oh scratch that, a lot of packages do have source set to #f

<vagrantc>i'm curious about submitting an update for a new version of a package ... being new to guix and all that, not sure what the norms are with new upstream versions ... wouldn't make any other packaging changes.

<vagrantc>specifically, i'd like to see u-boot version 2018.05 (currently at 2018.01, and missed 2018.03 entirely)

<lfam>vagrantc: In general, we like to follow upstream's recommendations. So, if we have the humanpower to do normal Guix packaging QA processes, and the update does not cause too many depending packages to be rebuilt, we do the update.

<lfam>So, if 2018.05 is the newest release (or at least newer than what we have), and it works for you, we should update it.

<vagrantc>lfam: e.g. feel free to submit version updates after some testing?

<lfam>" and the update does not cause too many depending packages to be rebuilt": This can be checked with `guix refresh --list-dependent foo`. We aim to cause <300 rebuilds for package updates on the master branch

<lfam>Yes, please feel free :)

<vagrantc>lfam: hmm... since u-boot isn't a public package itself, "guix refresh --list-dependent u-boot" just gives an error for an unknown package ...

<vagrantc>lfam: do i need to manually specify all the packages that inherit u-boot ?

<lfam>vagrantc: It's safe to guess that not many packages depend on u-boot, right?

<vagrantc>lfam: probably so...

<vagrantc>ACTION just wants to "do it right" :)

<lfam>Everyone appreciates that :)

<vagrantc>ACTION will build test u-boot-* and test a couple boards

<lfam>Reading the source of (gnu packages bootloaders), I see that the u-boot package isn't public. But you can still list its dependents like this: `guix refresh --list-dependent --expression='(@@ (gnu packages bootloaders) u-boot)'`

<vagrantc>aha

<lfam>It won't pick up the packages that inherit from u-boot via make-u-boot-package, but it's a good thing to know how to do in general

<vagrantcish>No dependents other than itself: u-boot@2018.05

<Apteryx_>vagrantcish: (I take you are the same person as vagrantc?) how can I make Guile & guile modules available in an SSH shell?

<Apteryx_>I have to source ~/.profile before Guile becomes available.

<Apteryx_>(on a foreign distro).

<Apteryx_>oh, I guess bashrc

<Apteryx_>(for non-interactive shells)

<Apteryx_>ugh. Didn't work.

<Apteryx_>(Ubuntu 16.04 is the foreign distro)

<vagrantcish>Apteryx_: i just followed the instructions in the guix manual

<vagrantcish>Apteryx_: but i'm mostly using guixsd, so i tend not to know how that stuff works

<thomassgn>good morning guix!

<thomassgn>anyone know how I could figure out if program X is part of my system? I would like to know if the xdg-user-dirs is installed and used by my system, but I can't see it in my path or anything...

<thomassgn>I see it's not even in my store, so it's not on my system.

<thomassgn>:)

<pkill9>thomassgn: `guix package -I` lists packages installed by the user

<thomassgn>pkill9: mm, but how about packages used by the system, "invisible" to the user?

<thomassgn>another question, the nix importer fails here because 'In execvp of nix-instantiate: No such file or directory'... <goes to check manual>

<roptat>"guix gc; find /gnu ..." maybe there's a better solution :)

<pkill9>thomassgn: `guix package -p /run/current-system/profile -I`

<pkill9>yeah i don't think the nix importer works

<pkill9>the nix package doesn't build either

<pkill9>and i think the nix importer depends on the nix daemon

<thomassgn>ah. dang

<thomassgn>well well. I'll try without then :)

<thomassgn>and then completely unrelated, anyone know of a description of daemon programming in C or maybe guile/scheme?

<pkill9>cool, the Hurd kernel is packaged in Guix

<soundtoxin>what kind of hardware support does hurd have? does it target any new stuff like arm64/openpower/risc-v or is it just playing catchup?

<pkill9>soundtoxin: not sure, I assume it has very little hardware support, however that's just my uneducated guess

<rekado_>soundtoxin: the hurd doesn’t have enough developers for targeting new stuff.

<rekado_>soundtoxin: but what’s great about that is that there’s a lot of space to express yourself in the project.

<bzp>hi all

<bzp>install openssh as configured? and I execute the service?

<roptat>bzp: on guixsd you need to add the service in your configuration and run guix system reconfigure

<roptat>then the service will be available (you can run "herd status ssh-daemon" to check)

<bzp>how do I configure and start up openssh

<roptat>you have to add something like "(service nginx-service-type (nginx-configuration))" in the services field of your operating-system declaration

<pkill9>ther'es an example somewhere in the guix source

<pkill9>gnu/system/examples

<roptat>look for openssh-service-type in https://www.gnu.org/software/guix/manual/html_node/Networking-Services.html for the configuration options you can set

<pkill9>~/.config/guix/latest/gnu/system/examples

<kkebreau>Has anyone else had Guix's substitute* procedure seemingly choke on an accented vowel in source code? At least I've found an instance where it does that.

<roptat>kkebreau: yes I have

<roptat>maybe your file is not proper UTF-8?

<kkebreau>roptat: It's an ISO-8859 encoded C++ file.

<kkebreau>Hm, looks like that might be the problem.

<roptat>kkebreau: (with-fluids ((%default-port-encoding "ISO-8859-1")) (substitute* …))

<kkebreau>roptat: Thanks! This fixes the issue!

<roptat>something is wrong with top... GiB Mem :113260510576640,0/15,532

<pkill9>how would you put the following string as the first argument to the substitute* function:

<pkill9>VARIABLE="${0##*/}"

<rekado_>"VARIABLE.*" ;)

<pkill9>ha :P

<rekado_>or: "VARIABLE=\\"\\\\$\\\\{0##\\\\*/\\\\}\\""

<efraim>you might need \\\\\\" for the quotation marks

<rekado_>no, just “\\"” is fine.

<rekado_>a string is enclosed in " ", so we just need to escape it once.

<rekado_>it has no meaning in a regular expression.

<rekado_>the $ and { and } need to be escaped in the context of the regular expression

<rekado_>but escaping them is done with \\ which has special meaning in Guile strings, so it needs to be escaped in the context of Guile once more.

IRC channel logs

2018-05-12.log