<toon`>wigust: tank you try it tomorrow <rekado>re ghc: it’s not just used for xmonad, but also for ghc-pandoc, which is used by r-rmarkdown, which is a very important R package. Since so much depends on pandoc, and pandoc depends on a lot of haskell packages I wanted to be sure that the ghc patch actually fixes our build problems <rekado>before merging core-updates into master. <rekado>I don’t think it will delay our plan to merge core-updates into master before this week ends. <pkill9>is there a package available for device-mapper? <pkill9>oh i think it's part of lvm2 which guix has <mbakke>Regarding EFI, I think there was a recent regression that required the EFI directory to be mounted directly under the host /boot - I'll try to fix that before the next release. ***danialbehzadi1 is now known as danialbehzadi
<myst_>hi guys I'm not much of a techie but i'd like to test guixSd <myst_>when installing, I will use an existing linux to controll grub, so how do i run the command --no-bootloader <myst_>so guixsd does NOT configure grub ***danialbehzadi1 is now known as danialbehzadi
***erdic_ is now known as erdic
<g_bor>I would be interested to use GuixSD in a project I have. <g_bor>This will be the first time I use it as a desktop. <g_bor>How can we package icecat addons/plugins? <g_bor>I also have another question. I use postgres a lot. Can we somehow provision users and databases, and maybe languages and extenions to databases? <OriansJ>g_bor: well as guix is written in guile, we could always extend it to encompass additional functionality if enough developers think it is important enough to implement. <OriansJ>g_bor: as for the packaging of icecat addons, I believe you can find that info with the existing icecat package <pkill9>temp_trisquel: Godot is in the Guix repositories, which is written in C and C++ <pkill9>it's a whole game development environment as well <pkill9>run `guix package -A godot` and you'll see it's there <rekado_>g_bor: there have been some discussions about the postgres service, but I don’t recall anything about database extensions. <rekado_>g_bor: it’s certainly possible, but it would require that someone investigates how this should be implemented. <taylan>can I use a UUID target for the bootloader part of the system config? <taylan>hmm, /dev/disk/by-id/... seems to come closest to having a persistent name for a disk <toon`>taylan: Maybe yes. It mentioned in documentation and in example of system configuration <taylan>toon`: where exactly? I don't see it. note that I mean the bootloader part, not the mapped-devices or file-systems parts <toon`>taylan: Sorry, missinterprete your question <toon`>taylan: what exacly you want uuid in target? <taylan>toon`: some sort of unique identifier... /dev/disk/by-id/ seems to come closest. the serial number is included so the worst thing that can happen is that it doesn't find the device because it was connected to a different interface. at least it won't write GRUB to the wrong disk :) <taylan>looks like EFI/GPT doesn't have this problem, because the bootloader is written to a partition and not the MBR of a disk. ***toon` is now known as toonp
***toonp is now known as toon_
***toon_ is now known as toon`
<efraim>more personal emails from Bruno, libffcall 2.1 has been released <efraim>hopefully that'll help with clisp on aarch64 on core-updates <lmao_>where can I learn to configure GuixSD? <OriansJ>lmao_: well we have example configurations in our manual which you could use <lmao_>the gnu manual doesn't feel very informative <lmao_>for example if I would want to disable slim and use stumpwm, I really don't know what should I do <OriansJ>lmao_: well slim is a login manager and a service that can be trivially removed from your definition file. Stumpwm is a program just like i3-wm and can be added in packages like any other program <pkill9>yeah the documentation could be improved <OriansJ>pkill9: well, we are hoping someone would be willing to take the time to help us improve our documentation from the perspective of an outsider and we would be more than happy to help prove them with any information they need along the way <lmao_>I really don't know what should I write to the file if I would want to change login manager <OriansJ>lmao_: well you could use %base-services and add the service associated with the login manager you desire or if you wish for alot of control, you can look into gnu/services/desktop.scm to see exactly what services are included in %desktop-services and replace anything there you don't like <efraim>with the bump to libffcall in master clisp still builds on aarch64 <efraim>sneek: later tell civodul I think the issue was / (with /tmp) was mounted with 'commit=600' and that was too slow for building <mbakke>rekado: Do you think core-updates will be merged today? <mbakke>I have some Python fixes, but don't want to push them to core-updates in case you're in the middle of a merge. <ng0>I just learned about: git archive --format=tar --prefix=PROJECT_NAME/ --remote=USER@SERVER:PROJECT_NAME.git master | tar -xf - (and special github case: curl -L https://api.github.com/repos/VENDOR/PROJECT/tarball | tar xzf - ) would this cost more CPU time than cloning and then removing the .git folder like we do currently? or, does guile-git support git archive? <ng0>this creates a tarball, and the tarball can be reused, saving a bit of space. just wild guesses, I never used git archive <OriansJ>efraim: depends if you use a commit or a tag id <OriansJ>In the first case the current time is used as the modification <OriansJ> time of each file in the archive. In the latter case the commit time as recorded in the referenced commit object is used instead. <OriansJ>The local and the github special case end up with different sha256sums in the testing I just did with mescc-tools <OriansJ>4282f2b981ef176230b60b98b43de0baf9a260287afe65de48082862a71c4f32 and f88b1785d26adc4b3a9aff2d246d52c2952c08cd166cd6e07bd57f7dc27ebda1 to be precise <OriansJ>but git archive Release_0.3 --format=tar >| trial does seem to generate consistent results locally <sneek>civodul, you have 1 message. <sneek>civodul, efraim says: I think the issue was / (with /tmp) was mounted with 'commit=600' and that was too slow for building <ng0>what's the magic I need to mumble to get fastcgi working on GuixSD? Nginx is already running <ng0>install fastcgi in the global profile? <civodul>we didn't get to cross-sign BTW? too bad <civodul>do you have a tool to generate this? <lfam>efraim: I wonder how you calculated it? Andreas Enge and I signed each other's keys <efraim>i'm working on turning it into a real package <efraim>I grabbed the keys from savannah, put them into a keyring and updated it from the keyservers <lfam>Perhaps Andreas or I forgot to upload the signature <efraim>the graph is from sig2dot | dot -Tpdf <ng0>oh, we have a service for it <lfam>The linter complains, "use @code or similar ornament instead of quotes" <lfam>I'm not sure what the solution is, or if I should do anything at all. Those names are not code <mbakke>lfam: replacing the single quotes with double quotes work. <mbakke>Maybe @dfn would be appropriate for some of those. <civodul>lfam: in theory @dfn or ``weird quotes'' would do <civodul>mbakke: so you're back from your road trip and the thing isn't merged yet! <civodul>but that's because the QA team is hard at work <civodul>ACTION attempts to fix a couple more packages on core-updates for good measure <mbakke>civodul: I'm actually still on the road, just having a "day off" :) <mbakke>Well not on the road, back in Prague at least. <civodul>mbakke: oh, nice to have a "day off" while on vacation :-) <mbakke>Prague is great, considering moving here actually :) <civodul>lfam: databases.scm, like memcached?