IRC channel logs

<lfam>rekado_: It didn't make a difference

<nckx>rekado_: I was going to ask about r-crosstalk, but see it's already been mentioned. Is there a reason it hasn't been pushed yet?

<rekado>nckx: what commit are you on? All the necessary fixes have already been pushed.

<efraim>rekado: r-plotly now has a dependancy on r-crosstalk

<janneke>morning guix!

<temp_trisquel>Hi.

<temp_trisquel>Is non-free data removed from Angband in Guix?

<temp_trisquel>Also, please update it if possible.

<solene>hello there

<temp_trisquel>I have forgot to use herd start cow-store while installation.

<temp_trisquel>Can I start it now?

<temp_trisquel>I have already seen "No space left p

<temp_trisquel>*on device

<temp_trisquel>"

<rekado>efraim, nckx sorry about this; something really got screwed up in my branch.

<rekado>I added r-crosstalk, which depends on r-shiny

<rekado>but after rebasing in a very convoluted manner I accidentally dropped r-crosstalk, and then concluded that r-shiny could wait after all.

<rekado>:-/

<rekado>so, I’m going to push r-shiny, and add r-crosstalk real quick.

<temp_trisquel>rekadl

<temp_trisquel>*rekado

<temp_trisquel>rekado: I have tried iomem=relaxed.

<temp_trisquel>OS has rebooted for unknown reasons and I was unable to boot it in some seconds so I decided to reinstall it.

<temp_trisquel>When I tried to mount ESP, it said partition is unmounted incorrectly.

<temp_trisquel>I need to try EFI installation again, last time.

<temp_trisquel>If it will fail, I will use BIOS.

<ng0>I have a problem with this: https://github.com/halcy/Mastodon.py/blob/d1b7d89eaa388d48ffb8e0b47ed023a6958e8823/setup.py . I already patched upstream, but it still fails, this time with a package we have and I provide. I'm no python expert. Could the "find_packages" be a proble mfor us here: from setuptools import setup, find_packages

<ng0>ah, the error message, one moment

<ng0> https://krosos.org/paste/f-pymastodon.txt

<ng0>As you can see here I provide python-dateutil https://gitlab.com/ng0_guix/guix/commit/53e4b99422e750368cbf0aeab20e65162d43b094

<ng0>the check phase seems to parse the dependencies.. and then goes: error: Could not find suitable distribution for Requirement.parse('dateutil')

<rekado>efraim, nckx it should be fixed now. I added r-crosstalk and r-shiny.

<rekado>the source of my confusion was git worktrees.

<rekado>I prepared the CRAN mass upgrades in a separate branch in a separate worktree, but applied the patches in a different worktree.

<rekado>I did some work in the wrong worktree, so my build tests simply didn’t provide the assurances I thought they would give me.

<jmi2k>Is the documentation for connman

<jmi2k>Is the documentation for connman wrong? It says: disable-vpn? (default: #f) When true, enable connman's vpn plugin.

<jmi2k> https://www.gnu.org/software/guix/manual/html_node/Networking-Services.html

<efraim>hmm, looks like i need to fix clisp on aarch64 to build javascript packages

<atw>sneek: later tell jonsger I found out the form factor of the 9360. It's M.2, but I don't know the key type. ref: http://topics-cdn.dell.com/pdf/xps-13-9360-laptop_setup%20guide_en-us.pdf . I also found https://wiki.archlinux.org/index.php/Dell_XPS_13_%289360%29 which indicates that on this specific model of the XPS 13, graphics is intel and wireless is ath10k. Support for both should be good. ath10k:

<atw> https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/tree/drivers/net/wireless/ath/ath10k . So, after looking into how to replace the card, I won't even have to do it :P

<sneek>Will do.

<ng0>isn't ath10k only in linux mainline and not libre?

<atw>jmi2k: http://git.savannah.gnu.org/cgit/guix.git/tree/gnu/services/networking.scm#n941 I think you're right; it does look wrong

<atw>ng0: I don't know for sure. I looked in the kernel source (see my third link) and the source is there. However, this doesn't necessarily mean it's free. Can I consult linux-libre documentation?

<ng0>2017-07-02 linux-libre news:

<ng0> Updated blob names for si2168 dvb, si2157 tuner, ath10k wifi, iwlwifi, rtl8723be wifi, rtl8821ae wifi, and silead touchscreen.

<ng0> https://www.fsfla.org/ikiwiki/selibre/linux-libre/

<ng0>ACTION sideeyes to the early and now old ath10k dongle

<atw>just found the same thing :/ well, I guess I can install a new card

<atw> https://www.fsfla.org/svn/fsfla/software/linux-libre/scripts/deblob-4.9 : "reject_firmware drivers/net/wireless/ath/ath10k/core.c" doesn't get any clearer than that

<atw>wish I could find the rationale for why it's nonfree though. https://wireless.wiki.kernel.org/en/users/drivers/ath10k doesn't mention anything about it. What license is https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/tree/drivers/net/wireless/ath/ath10k/core.c#n5 ?

<atw>s/about it/about nonfreedom/

<rekado>jmi2k: it’s confusing to use double negatives in configuration options. I’d rather have “enable-vpn?” with a default value.

<atw>rekado++

<solene>stupid question: how does sound work ? Do I have to start some pulseaudio service ?

<jmi2k>rekado: yes i agree. Also, the documentation is wrong: it says that when disable-vpn? is #t, it 'enables' vpn plugin.

<solene>I installed alsa-utils and I can control sound with alsamixer :)

<rekado>solene: often pulseaudio is started as needed.

<rekado>there is no system-wide service for pulseaudio; I believe it is dbus-activated and runs once per user session on demand.

<rekado>you can use alsamixer to control sound levels close to the hardware, but for applications that use pulseaudio you may want to use pavucontrol

<rekado>it gives you per-application volume control.

<solene>rekado: I don't think I have dbus installer or started. I know I should add it to my system configuration

<rekado>solene: are you using the default desktop services?

<solene>rekado: yes, I think, I have %desktop-services

<solene>and I see a dbus process with ps

<rekado>the desktop-services include dbus services.

<pmikkelsen>hi guix

<rekado>pmikkelsen: hi!

<rekado>hmm, on an installation of GuixSD I get this error when trying to install anything:

<rekado>guix package: error: unsupported manifest format

<rekado>oh, I bet I used a more recent version to build the current profile

<nckx>efraim, rekado: thanks!

<temp_trisquel>Hi.

<temp_trisquel>I forgot to execute herd start cow-store /mnt before installation.

<temp_trisquel>guix system init has downloaded some packages, then I pressed ^C and did herd start cow-store /mnt

<temp_trisquel>Will this cause errors?

<temp_trisquel>Help!

<rekado>your packages will have to be downloaded anew, because before cow-store was started they won’t appear on disk.

<temp_trisquel>Will they just be downloaded again?

<temp_trisquel>*be just

<rekado>I recommend rebooting.

<rekado>because I don’t know the details

<rekado>cow-store remounts the store, but I don’t know if it only applies to new writes

<temp_trisquel>So, broken system can be in result?

<temp_trisquel>guix system init is running. I don't see error message.

<temp_trisquel>Will my system be bootable/working (if bootloader will be installed correctly)?

<temp_trisquel>Help!

<rekado>temp_trisquel: you can try this by rebooting.

<temp_trisquel>There were some GRUB problems. It's possible my system will not boot because of broken GRUB.

<rekado>what GRUB problems?

<temp_trisquel>rekado: GRUB wasn't able to be installed correctly.

<rekado>please provide more details

<temp_trisquel>I'm trying with memio=relaxed now.

<temp_trisquel>If this will also fail, then Intel did EFI 'just for fun' here.

<rekado>before doing a full installation you could try to just install GRUB

<rekado>that might save you some time.

<temp_trisquel>Is non-free data removed from angband package in GuixSD?

<rekado>temp_trisquel: take a look at the package definition with “guix edit angband”

<rekado>temp_trisquel: you will see that the sources have a snippet attached that removes some directories.

<temp_trisquel>Is this removable data included in sources?

<rekado>temp_trisquel: when something is removed with a snippet it also affects the source tarball you get with “guix build -S angband”

<rekado>this tarball provided by Guix won’t include the removed data.

<temp_trisquel>What does GuixSD version mean? It's rolling release, I think.

<rekado>it’s the same as the version of Guix.

<lfam>With cuirass, is the best way to remove a job to use some SQL statement on the database?

<lfam>I'm working on getting cuirass set up on a foreign distro :)

<ng0>becasue we can't shallow clone yet, is 355 MB git checkout too big? I think it's okay. the checkout takes only a couple of minutes here. I was thinking of starting a package attempt of https://github.com/google/fonts and split them up into different outputs.

<ng0>or diufferent packages which inherit from each other, whjatever is smaller and easier

<paroneayea>boop

<bavier`>ng0: is might be possible to download individual fonts from fonts.google.com

<bavier`>s/is/it

<lfam>ng0: Probably we will never be able to shallow clone unless we use a Git library besides libgit2. They explicitly don't plan to support shallow clones

<ng0>bavier`: it kind of is possible, I had a hack somewhere

<ng0>but with git we can put at least versions to it

<ng0>which isn't possible with Google Noto already

<ng0>at least when we start separating the font

<ng0>and I'd rather have control than 'oops, there goes the clusterfuck of our armada of versionless fonts, outdated again'

<bavier`>right, versioning would be nice I guess

<lfam>Has anyone used Cuirass with a local repository? It complains with 'unsupported URL protocol' when I try something like "file:///home/leo/blah"

<bavier`>ng0: you could define an <origin> that multiple packages use, which just copy or symlink the relevant directories.

<bavier`>ng0: our git checkouts aren't compressed currently, so there'd be no uncompression penalty for having each package access the store directory

<ng0>oh.. wait. we can download directly from githib right

<ng0>one moment

<ng0>Wouldn#t it be easier to just define a couple of download sources for each font, for example: https://github.com/google/fonts/blob/6d3047828cd43bf4bcbb2f3f7810714b6238909c/ofl/abyssinicasil/AbyssinicaSIL-Regular.ttf etc

<ng0>some fonts are even just one file

<ng0>there are larger fonts which feature 6 files or more

<ng0>but a majority has 1 or 2 files only

<ng0>and most of them are rather sleepy in their state, only some get updated

<ng0>*updated very regulary

<bavier`>ng0: it seems like many of the fonts are just pulled from other locations, which are versioned separately. it might be good to package directly from the source

<ng0>okay, then I guess I'll write down a list of the sources

<ng0>or do you have more info on that already?

<ng0>I haven't lookedi nto their building tools so far

<bavier`>ng0: I've just noticed that many of the FONTLOG.txt files point to the upstream website

<bavier`>which have versioned downloads available for the specific font

<bavier`>e.g. ananymouspro/FONTLOG.txt points to http://www.ms-studio.com which has https://www.marksimonson.com/assets/content/fonts/AnonymousPro-1.002.zip linked

<ng0>let's hope more than 50% are versioned ;D

<ng0>ACTION uses a font which isn't versioned

<ng0>(at least not in the filename)

<ng0>ah

<ng0>wordcount: 909.

<ng0>hm..

<ng0>is anyone up to a link gathering game?

<lfam>We already have a package of Anonymous Pro

<bavier`>that was just an example

<lfam>Yes, I just wanted to point it out to save ng0 some time

<bavier`>:)

<ng0>I could start a thread on -devel with the ones we don't have and then add the links to the found upstreams

<bavier`>I guess what I'm saying is that the fonts.google.com aggregation is nice, but I think we'd rather use upstream where possible?

<ng0>yes

<lfam>We could always package them individually and then offer a fonts-google metapackage

<ng0>that would be of interest for things like hosting a google fonts alternative

<ng0>*could

<bavier`>yup

<ng0>ACTION starts the email

<ng0>I assume "apache" includes some fonts with apache specific features?

<ng0>they deon't appear in the ofl/ diurectory

<ng0>for example this (taking from a listing) is missing: 4.0K jsmathcmbx10/ 4.0K kranky/

<ng0>oh

<ng0>they sort by licenses

<ng0>how obvious

<ng0>I'd feel easier about this by using the patches mailinglist (and the tracker). Does someone disagree on this ~1000 lines email with the task for URL collection being on there, with an distinguishable Subject?

<ng0>*feel better

<ng0>ah, picked the wrong list afterall. Well.. email is email.

<lfam>Has anyone seen this error before when starting cuirass? 'fatal: repository '#f' does not exist;

<lfam>I'm passing 'git://git.savannah.gnu.org/guix.git' as the URL for the repo

<buenouanq>thumbnails are not being generated in gnome

<buenouanq>not just nautilus, missing in the imageviewer too

<buenouanq>quick search suggests this was widespread, but fixed a few months ago on debain and ubuntu

<ng0>what do you mean? I had thumbnails in the filebrowser

<buenouanq>ng0: I used to - They haven't come back since my reinstall.

<buenouanq>I waited a couple weeks to mention it figuring they just needed time to be generated or something.

<buenouanq>alas, still none

IRC channel logs

2017-08-04.log