<samueldr>
Mic92's nix-review doesn't seem to change GC_INITIAL_HEAP_SIZE
<gchristensen>
just export it, should be done
<LnL>
a "good" initial size depends on the memory of your machine
<samueldr>
yeah, it's already running with it; I kinda know what it's for, set it to 24G (64GB ram)
<samueldr>
but I'm out of my element since I kind of think something is wrong?
<gchristensen>
might be?
<Mic92_>
samueldr: what would have you expect there?
<samueldr>
Mic92_: your nix-review tool is fine, I was comparing to a run earlier this monday that should have shown the same issues I believe
<Mic92_>
And why would nix-review be able pick a better value then nix itself?
<v0latil3>
How do i set kernel module parameters in boot.kernelModules?
<samueldr>
Mic92_: read nothing into that, I was juste puzzled by the fact that it ran fine on monday and my nix-build with *probably* the same set of attributes doesn't right now
<v0latil3>
figured it out.
hellrazor has joined #nixos
<monokrome>
Does anyone know if there's a way to get secure boot to work w/ NixOS?
* monokrome
has no idea how secure boot works
<colemickens>
I'm actually surprised it isn't better integrated. I'd love to try to spitball it in a GitHub issue or something.
<clever>
the problem is that you must sign the kernel and initrd that are in /boot/
<monokrome>
Is that a money issue or is there some difficulty to signing it?
<clever>
and if your not using luks, an attacker can just read the secrets and then its pointless
<monokrome>
Sure, but I am using luks
<clever>
some firmware allow you to load your own key
<clever>
and even if you have luks, if the attacker does manage to defeat secureboot, he could boot your OS in a hypervisor, and then wait for you to enter the luks password
<clever>
secureboot doesnt stop your OS from being ran in a hypervisor
<monokrome>
yeah
<monokrome>
but they don't need to defeat secure boot if it isn't on in the first place
<clever>
so i could just swap out your motherboard, and boot your hdd under that
<clever>
oh, and have fun trying to sign the kernel modules in /nix/store/
<colemickens>
well, about hte MS leak to be exact ^
<kisik21>
clever: signing kernel modules is a thing I did at build-time when I was a gentoo user.
<clever>
kisik21: with the nix sandbox, you have to put the secret into /nix/store/
<clever>
kisik21: and now the secret is readable by every single user on the machine
<kisik21>
Oops
<clever>
hence "have fun" :P
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/fAWrT
<kisik21>
How about making a key and discarding it when everything is signed?
<kisik21>
no key, no leaks, no security breaches
<kisik21>
right?
<clever>
kisik21: your /boot changes every single time you `nixos-rebuild switch`
<clever>
kisik21: do you want to open your bios config and load a new public every time?
<kisik21>
Signing kernel modules, not the kernel itself
<clever>
then you will have seperate keys for the modules, bootloader, kernel, and config, potentially
<kisik21>
isn't that a good thing?
<gchristensen>
what is the purpose of a key which is never the same?
<clever>
yeah
<monokrome>
You wouldn't rotate keys w/ every switch
<gchristensen>
00:31 <kisik21> How about making a key and discarding it when everything is signed?
<samueldr>
this'll sound dumb, but where are the profile and stats shown?
<gchristensen>
samueldr: at the end
<samueldr>
using nix-build and --dry-run
<clever>
but some binaries are broken and dont show stats
<kisik21>
gchristensen: Yes, it's a bit strange, but kernel wouldn't mind. since modules are (I suppose) built at the same time and it always will have the latest key. And we discard the private part so nothing else can be ever signed.
<{^_^}>
nix#2391 (by grahamc, 8 seconds ago, open): Print stats in more programs
JonReed has quit [Ping timeout: 252 seconds]
Supersonic has quit [Ping timeout: 240 seconds]
alexteves has quit [Ping timeout: 246 seconds]
Supersonic has joined #nixos
alexteves has joined #nixos
<worldofpeace>
{^_^}:
<clever>
gchristensen: heh, that was a far simpler fix then i thought
<gchristensen>
:)
<clever>
i had patched it locally (and lost the diff), and it was editing more places, 2 or 3 i think
<clever>
let me find the other patch i have, it can probably be added to your pr
<{^_^}>
[nix] @grahamc closed pull request #970 → nix-collect-garbage: Add a note about rebuilding to clear /boot space → https://git.io/vK0IG
<samueldr>
grep texlive profile.txt | wc # makes me think it's possibly texlive combined with the amount of attributes that combine into busting even 10G
<clever>
samueldr: is nixops involved any?
<samueldr>
(gives 1805)
<samueldr>
clever: not at all
<clever>
samueldr: are you importing nixpkgs multiple times?
<samueldr>
I cleared my environment with `env -i` to remove the doubts coming from my overlay in the initial profiling
<Taneb>
Anyone know if there's a reason python matplotlib doesn't seem to be opening dialogs in a nix-shell? This is NixOS 18.03 running Gnome 3 and both python2 and python3
<srhb>
(Er, I guess, it is for me, but that's because I opt to have it that way)
<{^_^}>
[nixpkgs] @xeji merged pull request #45880 → qgis: patch for compatibility with cmake 3.12 → https://git.io/fAWYs
<{^_^}>
[nixpkgs] @xeji pushed commit from @thefloweringash to staging « qgis: patch for compatibility with cmake 3.12 (#45880) »: https://git.io/fAW9o
<Lisanna>
I mean, nix-channel doesn't seem to really do anything special anymore
<Lisanna>
it was useful before because it would download the nix expressions and put them somewhere, so your NIX_PATH could point to a fixed location
<Lisanna>
but since you can put URIs in your NIX_PATH now, it doesn't seem to really have any value
<Lisanna>
and probably just adds confusion especially to new users
<srhb>
Ah, I see. Well, yes, I sort of agree that it is confusing to new users and that explicitly pointing at channels is clearer.
* emily
sort of wishes someone would write a "setting up NixOS the Modern Way" post since all the documentation points to using the old busted methods of doing things
<srhb>
I think opinions differ on what the right approach is, but more resources doesn't hurt :)
<Myrl-saki>
How is nixUnstable suffix "calculated"?
<Myrl-saki>
Oh wait.
<Myrl-saki>
Number of commits?
<srhb>
Myrl-saki: The last part is the rev at least.
<Myrl-saki>
srhb: Yeah, I think the first value is number of commits. That's pretty funny.
<srhb>
That sounds right
LnL has quit [Read error: Connection reset by peer]
LnL has joined #nixos
georges-duperon has joined #nixos
thc202 has joined #nixos
Thra11 has joined #nixos
dmj` has quit [Quit: Connection closed for inactivity]
phreedom has joined #nixos
<Lisanna>
emily I mean, the right way to setup any kind of linux distro is going to be pretty subjective. There's only a little bit of boilerplate in the steps - most of it really does come down to stuff like "how do you prefer to partition your disk?" "UEFI or Legacy?" "Swap or no swap?"
johanot has joined #nixos
<emily>
Lisanna: sure, but I'm thinking specifically stuff like NIX_PATH etc., the "nix ..." tools of nix 2.0, tracking your configs in git...
<Lisanna>
ah, okay
<emily>
I dunno, it just feels like in a lot of instances I've ended up thinking "oh, that's how you do it". like keeping my own nixpkgs branch
marzzzello has quit [Quit: bb]
<jasongrossman>
IMO, the Nix/NixOS documentation is not *terrible*. It would be good to collect examples of projects that have better documentation, and which are also fast-developing projects like NixOS, and are non-commpercial projects like NixOS.
<emily>
well, citing Rust would be cheating
kisik21 has quit [Ping timeout: 252 seconds]
<emily>
since they have way more people
<jasongrossman>
Oh, and also which are multi-person projects. I particularly like the documentation for TeX, but of course Knuth wrote that.
<jasongrossman>
Mm, instead of "commercial" I should have said "well funded".
<jasongrossman>
Maybe all we need is a small team of people putting more work into the manual. Then other people could ad stuff to the wiki in a fairly ad hoc way, and the small team could move it into the manual. But I doubt that anyone wants to take on this job.
<jasongrossman>
* add
<jasongrossman>
That's how Emacs documentation works. The manual is considered a major project, and it incorporates material from all over the place including a wiki.
<jasongrossman>
Emacs doesn't change as fast as Nix though.
<{^_^}>
[nixpkgs] @FRidh opened pull request #45898 → pythonPackages.pytest: fix .pytest_cache, again → https://git.io/fAWQZ
<Havvy>
emily: Still, wouldn't hurt to use the tools we use. E.g. `mdbook` is pretty great.
<jasongrossman>
I imagine that what it needs to make this sort of thing happen is for one of the core developers to take it on, although not on their own.
<jasongrossman>
Because if someone like me offered to do it, I'd just be too ignorant about what the contents should be (and therefore the semantic organisation etc.).
<Havvy>
Asking for a basic outline and then filling it in would work well.
<jasongrossman>
Absolutely, but it's also a lot of work both to get the basic outline right and to organise people to fill it in ... although I think it would be fun work.
bennofs[m] has quit [Remote host closed the connection]
timclassic has quit [Remote host closed the connection]
ejpcmac has quit [Read error: Connection reset by peer]
MarkOtaris has quit [Remote host closed the connection]
simbergm has quit [Read error: Connection reset by peer]
Ralith has quit [Read error: Connection reset by peer]
Purple-mx has quit [Remote host closed the connection]
schmittlauch[m] has quit [Write error: Connection reset by peer]
tokudan[m] has quit [Remote host closed the connection]
fgaz has quit [Remote host closed the connection]
koschei[m] has quit [Remote host closed the connection]
roberth has quit [Remote host closed the connection]
bachp has quit [Remote host closed the connection]
florianjacob has quit [Write error: Connection reset by peer]
thefloweringash has quit [Read error: Connection reset by peer]
<jasongrossman>
Also, IMO the manual is already really really good, and the problem is to keep it up to date and make it more complete.
<jasongrossman>
(I guess I mean manuals plural.)
vcunat has quit [Ping timeout: 252 seconds]
cryptix has joined #nixos
<Havvy>
That manual tries to be all things for all people. I've been left asking questions looking at it more than a few times.
<jasongrossman>
Don't you think it could achieve that aim with a bit more work?
<Havvy>
I don't think _any_ documentation source can be all things for all people. You'll notice we have multiple books in the Rust community. (And we're gaining a lot of new ones this year as well)
jasongrossman has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @romildo opened pull request #45900 → dde-qt-dbus-factory: init at 1.0.4 → https://git.io/fAWdg
<teto>
any guide on how to enable coredumps ? I have ulimit unlimited and /proc/sys/kernel/core_pattern set to "core". Do you guys use systemd-coredump instead ?
kiloreux has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @dotlambda pushed to master « byobu: use python3 »: https://git.io/fAWdD
<Myrl-saki>
It's been so long since I've ran `sudo git checkout origin/nixos-unstale`
<Myrl-saki>
unstable*
<Myrl-saki>
whoops
<Myrl-saki>
OTOH, that actually half makes sense. :P
<teto>
Dezgeg: do you have any specific config ? I've enabled it yet /var/lib/systemd/coredump remains empty after a program crashes (there is an entry in journal)
<Dezgeg>
it should just be systemd.coredump.enable = true;
<teto>
systemd.coredump.enable = true; is not enough it seems
<teto>
ha didn't see the "Resource limits disable core dumping for process 30855 "
<Ashy>
uh oh, nix daemon ran out of memory
<Ashy>
8gb ram on this machine but no swap
<hodapp>
aaack
acarrico has joined #nixos
<hodapp>
I should put an SSD in my work laptop sometime :-/ the first fun of things like nix-env after a reboot can take ages
<{^_^}>
[nixpkgs] @xeji pushed commit from @romildo to master « dde-qt-dbus-factory: init at 1.0.4 (#45900) »: https://git.io/fAWFN
vcunat has quit [Ping timeout: 252 seconds]
theunknownxy has joined #nixos
<{^_^}>
[nix] @edolstra closed pull request #135 → Updated instructions and scripts for non-NixOS multi-user installations → https://git.io/fAWbt
<Ashy>
hodapp: im on my "new" T450 now (arrived yesterday)
<Ashy>
$300 on ebay :D
<hodapp>
looks like a decent price
<hodapp>
mine's a T410
<hodapp>
what's OpenGL support like on yours? if you've had any chance to use it
<Ashy>
anyway, a 250gb ssd is really quite cheap these days, i bought a 500gb samsung 960 a few months back because i noticed greping the linux kernel source was noticable on a sata ssd
<hodapp>
my desktop has had a nice NVMe SSD for awhile now and it's a huge help
<hodapp>
but it's only this year that I started doing a lot of real(tm) work on my laptop
<symphorien>
hodapp: I have a T420 and I have never experienced any particular opengl issue. Do you refer to something specific ?
<hodapp>
symphorien: mine is Intel-based graphics and some months back an update did something that gets it into a mode where it will render the first frame of an OpenGL scene and then it will never update
<hodapp>
so some things like Blender - which do their entire UI with GL - are completely unusable
<hodapp>
on a fresh reboot, it doesn't do this
<hodapp>
but I've not figured out what exactly causes it - some combination of suspending and of me running xrandr (I have it driving two LCDs and have a script to set it up the right way)
<symphorien>
glxgears works fine here
<hodapp>
Intel graphics?
<symphorien>
i915 yes
<teto>
Dezgeg: wow coredumpctl debug is cool. Thanks :D
<symphorien>
the only i915 related problem I had is a bug with xfce-power-manager and hibernation (black screen on resume) which I experienced only on linux 4.14 (not 4.9 nor 4.17)
vcunat has joined #nixos
<symphorien>
also you have to to early kernel mode setting to have graphics when using the crashdump kernel feature but it is quite niche
<samueldr>
well... that was easy... initially thought there would be gotchas sue to i686 on x86_64
<samueldr>
due to*
<teto>
I nix-shell this : `let devPkg = pkgs.haskellPackages.developPackage {...} in pkgs.haskellPackages.ghcWithPackages (p: with p; [hdevtools hlint devPkg])` to provide hdevtools/hlint tools to my editor. But now haskell can't import haskell dependincies pulled by devPkg. How should I write this ?
ma27 has quit [Quit: WeeChat 2.1]
<manveru>
hm, how does one install nixos then these days?
<manveru>
also, fdisk doesn't seem to have the 'a' command
<samueldr>
manveru: I use `cfdisk` when paritioning
<{^_^}>
#45345 (by vmandela, 1 week ago, open): nixos/install-grub: include child configs in grub menu
<samueldr>
manveru: in a way, but not necessarily locked-in
<manveru>
p_l: me :)
<samueldr>
p_l: I, too
<manveru>
samueldr: i'll try those instructions then, but no clue how to setup my bios to use uefi...
<pie__>
how do I gc user profiles?
<vmandela_>
I am pulling in the test I wrote from a separate file into installer.nix. I need to include some additional configuration and a few extra test steps.
<p_l>
samueldr, manveru: Can you share your approach somewhere? I got to the point of running, but it's annoying a bit
<symphorien>
manveru: iirc uefi => gpt but legacy supports gpt
<p_l>
(using nix-shell currently)
<samueldr>
manveru: one quick trick, `dmesg | grep 'efi:'` if you have messages from the kernel around 0.000000 you're booted using UEFI
<manveru>
p_l: nix-shell, direnv, bundlerEnv
<samueldr>
(also presence of efivarfs in mount)
<vmandela_>
I am able to use the extraConfig option to pass the configuration I need into the tests. I am running into string escaping issues when trying to include tests.
<manveru>
samueldr: nothing there
<samueldr>
p_l: nix-shell only for development
<samueldr>
manveru: you're probably booted using legacy or mbr I guess then
<manveru>
i'd guess so
<manveru>
lemme see how i can wipe that gpt table
<samueldr>
do you know if the installer booted using `isolinux`? if so, it's definitely legacy boot
<manveru>
no clue
<samueldr>
(the installer image probably should telegraph the boot method somehow to the end-user, I'm thinking a kernel command line param that can be used in a diagnostic script)
<manveru>
ah, gdisk seems to have an option to convert back to MBR
<p_l>
samueldr: any examples tolook over?
<samueldr>
not really, for developement I use a nix-shell, which has `ruby` in it (bringing in bundler) adding all other native dependencies needed (e.g libiconv for nokogiri)
<p_l>
manveru: what exactly are you doing with direnv and bundlerEnv?
<mconstant>
I'm also interested in using Rubymine with direnv,nix-shell,bundix, etc
mayhewluke has quit [Ping timeout: 252 seconds]
mayhewluke has joined #nixos
aexoxea has quit [Quit: Goodbye!]
<siers>
How should I put my little nix-gitignore project in nixpkgs? Preferably with fetchgit or rather by copying it directly into nixpkgs? https://github.com/siers/nix-gitignore/issues/6
<{^_^}>
siers/nix-gitignore#6 (by CMCDragonkai, 1 day ago, open): Integration into Nix/Nixpkgs
<gchristensen>
builtins.fetchgit can't be used in nixpkgs, and nor can import pkgs.fetchGit
<p_l>
mconstant: for now ruby-mine starts for me in nix-shell, though I do get a bit of a mess with SDKs
<p_l>
now to get sql server running as well...
<p_l>
(I of course forgot to include freetds in shell.nix)
aarvar has quit [Ping timeout: 245 seconds]
fendor has quit [Read error: Connection reset by peer]
endformationage has joined #nixos
georges-duperon has quit [Ping timeout: 246 seconds]
<siers>
gchristensen, they mustn't be used in package definitions? Well fetchFromGitHub would also be fine
<worldofpeace>
Anyone here that would like to share their preferred debug process for nixos modules?
<clever>
worldofpeace: first step i usually do, is just nixos-rebuild build, and read the result to confirm its doing the right thing
<teto>
worldofpeace: nixos-rebuild has some options to run faster, read its help
<teto>
isn't ghcWithPackages supposed to provide ghc/ghci ? (nixos-unstable)
<clever>
teto: yes
<clever>
teto: how are you using it?
<teto>
I run nix-shell shell-haskell.nix with `let pkgs = import <nixpkgs> {}; in pkgs.haskellPackages.ghcWithPackages(p: with p; [netlink hlint hdevtools])`
<clever>
teto: that drops you into a shell suitable for building ghcWithPackages, not using ghcWithPackages
<clever>
teto: your shell.nix must return another derivation, that has that ghc in the buildInputs
JonReed has joined #nixos
<{^_^}>
[nixpkgs] @samueldr opened pull request #45910 → Fixes i686 installer iso build. → https://git.io/fAlJz
<{^_^}>
[nixos-homepage] @grahamc opened pull request #238 → Use nix-build to build the site → https://git.io/fAlJg
<aminechikhaoui>
manveru: I think any failure during the boot results in a kernel panic, not sure why we're setting that by default, I would personnaly prefer having dropping me into a shell instead
<samueldr>
the kernel will panic when PID 1 exits; though there are options for nixos (kernel command line options) to instead go to a shell
<gchristensen>
not nice to (by default) drop an unauthenticated user in to a shell as root with a possibly decrypted device
<gchristensen>
(iirc, that is a possible issue)
<hodapp>
are there any good examples of handling code samples (e.g. example plugins) that sometimes are present in repos?
<hodapp>
whether that is handling them as part of the main derivation for the repo, or having a separate derivation
<aminechikhaoui>
gchristensen: ah right
ambro718 has joined #nixos
<{^_^}>
[nixpkgs] @samueldr opened pull request #45911 → nixos/manual: nixos-help knows about colon-separated BROWSER → https://git.io/fAlTs
drakonis__ has joined #nixos
<hodapp>
but hmmmm. I want to make it easy to create a dev environment to develop Appleseed plugins in... but rather than having anything separate for this, wouldn't it just be preferable that if you "nix-shell -p appleseed" you have all the necessary libs/includes present?
<Dezgeg>
it is required for creating such packages but not to depend on such packages
<hodapp>
what do you mean?
<Dezgeg>
if you are writing a package foo that will depend on say, openssl (which uses multiple outputs), you don't need to make foo itself multiple-output
<p_l>
okay, I'm stuck. I'm trying to get rubymine to somehow see things as they are inside a nix-shell environment, and failing to do so. In this case, RubyMine is installed inside the environment and started in it
tenten8401 has joined #nixos
<tenten8401>
hey, so I recently got an RX 580 and I was trying to use it with NixOS, but it seems like my OpenGL version is stuck at 2.1 and none of my games launch... would anyone have any suggestions on what to do to fix this?
tv has joined #nixos
<tenten8401>
I've tried services.xserver.videoDrivers = [ "amdgpu" ]; hardware.opengl.enable = true; hardware.opengl.driSupport = true; and hardware.opengl.driSupport32Bit = true; none of which seem to be making any difference
johanot has joined #nixos
<{^_^}>
[nixpkgs] @vcunat pushed to master « nixos-generate-config: fix perl quoting after cb1ca42 »: https://git.io/fAltT
tenten8401 has quit [Quit: Konversation terminated!]
bachp has quit [Remote host closed the connection]
Purple-mx has quit [Remote host closed the connection]
tokudan[m] has quit [Remote host closed the connection]
schmittlauch[m] has quit [Read error: Connection reset by peer]
florianjacob has quit [Write error: Connection reset by peer]
thefloweringash has quit [Write error: Connection reset by peer]
timclassic has quit [Remote host closed the connection]
koschei[m] has quit [Remote host closed the connection]
simbergm has quit [Remote host closed the connection]
ejpcmac has quit [Read error: Connection reset by peer]
bennofs[m] has quit [Read error: Connection reset by peer]
roberth has quit [Read error: Connection reset by peer]
fgaz has quit [Remote host closed the connection]
MarkOtaris has quit [Remote host closed the connection]
Ralith has quit [Read error: Connection reset by peer]
kisik21 has quit [Ping timeout: 252 seconds]
<infinisil>
What's the difference between nix-deploy and nixos-rebuild --target-host?
<gchristensen>
you can get nix-deploy without nixos
<infinisil>
Hmm I see, I'll use nixos-rebuild thin
<infinisil>
then*
<infinisil>
Or try at least, experimenting with switching from Nixops
<dhess>
Anyone here running a Hydra on an EC2 t2 or t3 instance?
<ixxie>
infinisil: doesn't nix-deploy also support multiple targets?
<vcunat>
hydra.nixos.org uses some Amazon spot instances, I think, but I don't really know anything about that...
<{^_^}>
[nixpkgs] @xeji pushed to master « nixos-generate-config.pl: remove executable bit »: https://git.io/fAlto
<dhess>
vcunat: yeah I believe that's for dynamically provisioning remote builders, which is cool and I plan to use that. What I mean here is running the actual Hydra service on a t2 or t3.
periklis has joined #nixos
<dhess>
I'm wondering whether it's burstable enough for one of those or if it would just chew up too much CPU
<dhess>
obviously this depends on the number of projects etc. Just curious if anyone is actually doing it.
<gchristensen>
sure I did it
<gchristensen>
its fine
<gchristensen>
as long as you're not building nixpkgs a lot
<dhess>
gchristensen: wow cool! Which type did you use?
Lisanna has quit [Quit: Lisanna]
<gchristensen>
I forget. it was pretty cehap.
<dhess>
I was looking at the t3.medium
<dhess>
I'm a little worried about the 4GB RAM on that one
<teej>
For `nix-env -iA nixos.go` I receive `error: attribute 'nixos' in selection path 'nixos.go' not found`,
<symphorien>
replace nixos by nixpkgs
<symphorien>
are you on nixos or do you use nix on something else ?
<teej>
Oh I can use nixpkgs.
<symphorien>
for some legacy reason, nixpkgs is called nixos on nixos
<symphorien>
so nixos users will mislead you :)
<teej>
I'm using Nix on a Linux environment.
<vcunat>
the attributes are call the way how you call your channels
<vcunat>
*called
<teej>
So even when using `nix-env -iA nixpkgs.go` I received `error: out of memory`.
<vcunat>
well, that's weird, unless you do have very little memory :-)
<clever>
teej: what does it print before that, and how much ram do you have?
baimafeima has joined #nixos
<witchof0x20>
How would I go about adding additional vim plugins? I know there's nixpkgs/pkgs/misc/vim-plugins/vim-plugin-names and pkgs/misc/vim-plugins/default.nix
<witchof0x20>
Are the derivations in pkgs/misc/vim-plugins/default.nix auto-generated or hand-written?
<teej>
So before reaching `error: out of memory`, it says... `installing 'go-1.11' `these paths will be fetched (191.28 MiB download, 939.96 MiB unpacked): /nix/store/dana4hagc62f8ylf57rci90d5frgzigk-go-1.11` `copying path '/nix/store/dana4hagc62f8ylf57rci90d5frgzigk-go-1.11' from 'https://cache.nixos.org' ...`.
<{^_^}>
[nix] @edolstra merged pull request #2388 → Document multi-user installation, and add release notes about it not being the default on 2.1 → https://git.io/fACAq
<yosemitesam>
Is there a way to configure nix to clone a git repo and copy the whole content to some place in the filesystem? I know I can do a simple fetchgit, but how can I then copy the whole repo somewhere?
<clever>
yosemitesam: why do you need the repo copied somewhere?
<yosemitesam>
I want to keep a git repo with all my configurations and have it lie in my home directory. In the nix configuration I then want to create links to these files. This way I have the git repo on every machine and can simply update a file there and have the changes be live. If the change is good, I can git commit&push to get it on my other machines...
<{^_^}>
[nixpkgs] @xeji pushed commit from @greydot to master « qcomicbook: init at 0.9.1. (#44074) »: https://git.io/fAl3L
<clever>
for example, i want to define options in one place, config in another, then based on the config, define config for machines
<witchof0x20>
JonReed: thanks! I think i'm using whatever home-manager uses, but i've had vim-configurable in the past, and this looks fairly simple
<gchristensen>
yay so many Nix PRs merged today
<gchristensen>
niksnut++
<{^_^}>
niksnut's karma got increased to 2
<shreyansh_k>
Hi guys, I'm trying to install kdeconnect and I added it into the systemPackages variable but I can't find the icon in the launcher to launch it.
<{^_^}>
[nixpkgs] @michaelpj opened pull request #45920 → freedesktop modules: disable by default → https://git.io/fAl3G
<shreyansh_k>
I'm using kde plasma
<shreyansh_k>
I've heard kde applications have to be installed using some wrapper code but I'm unable to find out how to go about it.
elvishjerricco has joined #nixos
<shreyansh_k>
Any help would be appreciated. Thank you.
Thra11 has quit [Ping timeout: 246 seconds]
<infinisil>
clever: pretty sure what I got here can do that
<infinisil>
It even allows different machines to use different nixpkgs versions!
<infinisil>
Which is another major negative point of nixops
<clever>
infinisil: nice
<clever>
infinisil: one thought on what you could now do, mystuff.loadbalancer.backends = 5;
<clever>
infinisil: and then spin up 5 backends, for a single loadbalancer, a total of 6 machines
<clever>
currently, nixops has no way for config in 1 file to cause another file to create machines
<infinisil>
clever: you could do it on a non-module system level
<infinisil>
But yeah, it's much more flexible when it's via module system
<clever>
yep
<elvishjerricco>
infinisil: Can you set `nixpkgs.pkgs` in a nixos module to get each machine using a different nixops?
<clever>
elvishjerricco: part of the problem, is that that option only effects the pkgs tree, not the nixos tree
<infinisil>
elvishjerricco: Not within the nixos module, but rather on a level higher up
<clever>
elvishjerricco: it is the nixos modules that are responsible for reading the config and obeying nixpkgs.pkgs, and then its too late to change the nixos modules
oida has quit [Ping timeout: 250 seconds]
<infinisil>
Yeah, so it's these "nixops level modules" that can then allow you to set (what I have now): { machines.foo.nixpkgs = /path/to/nixpkgs; }
<infinisil>
While the nixos config for foo would be in { machines.foo.config = { pkgs, ... }: { services.xserver.enable = true; }; }
Mateon3 has joined #nixos
oida has joined #nixos
<elvishjerricco>
Ah, yea ok
<elvishjerricco>
Nixops has a lot of problems. I haven't really been thrilled with it
<infinisil>
elvishjerricco: Are there any others than the ones we mentioned you ran into?
<elvishjerricco>
infinisil: Probably nothing that's not in the issue tracker.
<infinisil>
Was hoping for something more specific :P
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon3 is now known as Mateon1
<elvishjerricco>
infinisil: The duplicated eval, the phases having an effect on when certain things can be used, the state not being encrypt-able, hard coded paths in the state file, etc..
<elvishjerricco>
My pettiest issue with it is that it doesn't use the Nix 2.0 one-line output :P
<shreyansh_k>
Hi, anyone uses kde5 here?
<infinisil>
elvishjerricco: What's the phases thing exactly?
<shreyansh_k>
newbie here
<infinisil>
,ask shreyansh_k
<{^_^}>
shreyansh_k: Just ask your question. It's the best way to know if anybody can help.
<shreyansh_k>
Cannot edit network configuration from the desktop... "Could not find module 'kcm_networkmanagement.desktop'. See kcmshell5 --list for the full list of modules."
<elvishjerricco>
For instance, SSH keypairs can't be deployed to deployment.keys
<elvishjerricco>
Basically, something defined in phase 1 can't be used in phase 1. Really it needs to be some sort of monadic thing, but I dunno how to do that without rewriting in C++ so that we can create "impure" thunks.
logzet has quit [Ping timeout: 240 seconds]
<elvishjerricco>
I'd probably prefer a model where Nix expressions are used to generate scripts that perform deployment depending on a state file.
<infinisil>
Darn, I was a bit too early with my joy on the nixos-as-a-submodule thing
<infinisil>
Because imports don't work with submodules, and I've been bashing my head against the module system for the past hour to get it working, but no success so far
<infinisil>
I mean it does somewhat work, it's just a bit ugly
<infinisil>
Too ugly
baimafeima has quit [Quit: Leaving]
<clever>
infinisil: the requires flag in nixops is also not well known, and ive seen things get pretty messy without it
Ariakenom has quit [Ping timeout: 272 seconds]
<infinisil>
clever: What does that do?
<clever>
infinisil: mostly the same as imports, but without the fixpoint
<infinisil>
clever: Ah, so that's about the same as just specifying multiple nixops files?
<clever>
infinisil: yeah
<clever>
but you can make a tree, and manage that list in files, and not need to re-run `nixops modify` every time you pull
<infinisil>
Yeah
puffnfresh has joined #nixos
justbeingglad has joined #nixos
justbeingglad has left #nixos [#nixos]
dhess has quit [Remote host closed the connection]
Ariakenom has joined #nixos
<clever>
infinisil: and in the example i linked, there is some rather complex haskell code, that decides what goes into `nixops modify` based on several yaml files
<clever>
infinisil: but, i know that if the -staging file is in, i must also have the bucket storage one
<infinisil>
submodules that have an option named "config" only don't work with imports..
<infinisil>
I know why, but this sucks
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<infinisil>
I'll try to make it work though
<das_j>
How do I make nix not render newlines in strings? I have "mystring myotherstring" and I want both to be separated in the nix file but not in the resulting string
<clever>
das_j: [ "foo" "bar" ] ?
<clever>
if the newline is not within the quotes, nix basically ignores it
<das_j>
clever: But that's not a single string
<clever>
derivations will join arrays into strings automatically
<das_j>
oohh
<clever>
there are also functions to join arrays with a separator
<clever>
buildInputs is an example of an array turning into a space-separated string
<das_j>
I'll do that, thanks!
<das_j>
concatStringsSep is what I usually use
<clever>
yeah, thats one
<das_j>
But I thought there was a way that is built into the language