charego has quit [Remote host closed the connection]
shabius_ has quit [Remote host closed the connection]
<clever>
,libraries mmchen
<{^_^}>
mmchen: Don't install libraries through nix-env or systemPackages, use nix-shell instead. See https://nixos.wiki/wiki/FAQ/Libraries for details.
shabius_ has joined #nixos
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
KindOne has quit [Quit: K-Lined]
orivej has joined #nixos
<mmchen>
clever: what's weird is that it compiles fine but the lsp server doesn't seem to recognize the header
NobbZ[m] has quit [Ping timeout: 246 seconds]
NobbZ[m] has joined #nixos
ngerstle[m] has quit [Ping timeout: 246 seconds]
limpid-kzonix[m] has quit [Ping timeout: 246 seconds]
citizenzibb[m] has quit [Ping timeout: 246 seconds]
adit[m] has quit [Ping timeout: 246 seconds]
wedens[m] has quit [Ping timeout: 246 seconds]
ksevelyar[m] has quit [Ping timeout: 246 seconds]
dtz has quit [Ping timeout: 246 seconds]
metheflea has quit [Ping timeout: 246 seconds]
leward[m] has quit [Ping timeout: 246 seconds]
cellofhuman[m] has quit [Ping timeout: 246 seconds]
piotrszegda[m] has quit [Ping timeout: 246 seconds]
loupinfre[m] has quit [Ping timeout: 246 seconds]
hsngrmpf[m] has quit [Ping timeout: 246 seconds]
d4rkshad0w has quit [Ping timeout: 246 seconds]
wildtrees[m] has quit [Ping timeout: 246 seconds]
rfglinux[m] has quit [Ping timeout: 246 seconds]
Luigi[m] has quit [Ping timeout: 246 seconds]
antpic01[m] has quit [Ping timeout: 246 seconds]
counterfactual[m has quit [Ping timeout: 246 seconds]
Jonas[m]3 has quit [Ping timeout: 246 seconds]
Criena has quit [Ping timeout: 246 seconds]
faya[m] has quit [Ping timeout: 246 seconds]
lycium[m] has quit [Ping timeout: 246 seconds]
morr[m] has quit [Ping timeout: 246 seconds]
PkmX[m] has quit [Ping timeout: 246 seconds]
f0i[m] has quit [Ping timeout: 246 seconds]
virtusviri7321[m has quit [Ping timeout: 246 seconds]
tbenst[m] has quit [Ping timeout: 246 seconds]
dittofan[m] has quit [Ping timeout: 246 seconds]
anarchomoh[m] has quit [Ping timeout: 246 seconds]
dpc has quit [Ping timeout: 246 seconds]
mt[m] has quit [Ping timeout: 246 seconds]
hsiktas[m] has quit [Ping timeout: 246 seconds]
dngray has quit [Ping timeout: 246 seconds]
<clever>
mmchen: is the lsp server running under nix-shell?
rycee has quit [Ping timeout: 265 seconds]
ggpeti[m] has quit [Ping timeout: 265 seconds]
chrismatheson has quit [Ping timeout: 265 seconds]
tamwile[m] has quit [Ping timeout: 265 seconds]
mkg20001 has quit [Ping timeout: 265 seconds]
axx[m] has quit [Ping timeout: 265 seconds]
Yakulu[m] has quit [Ping timeout: 246 seconds]
aparkerdavid has quit [Ping timeout: 246 seconds]
comrandroxaos[m] has quit [Ping timeout: 246 seconds]
gcoakes[m] has quit [Ping timeout: 265 seconds]
kazzimazzi[m] has quit [Ping timeout: 265 seconds]
kozova1[m] has quit [Ping timeout: 265 seconds]
srid has quit [Ping timeout: 265 seconds]
gueorgui has quit [Ping timeout: 272 seconds]
puzzlewolf has quit [Ping timeout: 246 seconds]
maralorn has quit [Ping timeout: 246 seconds]
<energizer>
i see anbox is in nixpkgs. is it possible to run android apps with it in nixos?
matthewcroughan has quit [Quit: Connection closed]
<clever>
energizer: i believe mobile-nixos is already able to do that as well
pamplemousse has quit [Ping timeout: 264 seconds]
ngerstle[m] has joined #nixos
virus_dave has quit [Quit: virus_dave]
tamwile[m] has joined #nixos
<energizer>
clever: i mean in regular nixos though
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JfrAy
<gchristensen>
energizer: I don't think it can, I think running android apps is out of scope for now
gueorgui has joined #nixos
<energizer>
alright
ggpeti[m] has joined #nixos
kozova1[m] has joined #nixos
<gchristensen>
oops
rycee has joined #nixos
gcoakes[m] has joined #nixos
chrismatheson has joined #nixos
<gchristensen>
sorry that was about mobile-nixos
srid has joined #nixos
kazzimazzi[m] has joined #nixos
KindOne has joined #nixos
mkg20001 has joined #nixos
axx[m] has joined #nixos
maralorn has joined #nixos
dngray has joined #nixos
mt[m] has joined #nixos
dittofan[m] has joined #nixos
hsiktas[m] has joined #nixos
<jtojnar>
clever turns out I cannot just use flake inputs in my configuration (makes sense since I do not pass them to eval-config.nix)
leward[m] has joined #nixos
pamplemousse has joined #nixos
<bqv>
energizer: I've done it
d4rkshad0w has joined #nixos
ksevelyar[m] has joined #nixos
cellofhuman[m] has joined #nixos
<jtojnar>
I need to pass it `specialArgs = { inherit (inputs) nixpkgs; };`
<mmchen>
:clever i think so? the nix-shell has clang-tools which the dev environment runs it
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
<mmchen>
right now i just have `nix-shell -p gcc clang-tools`. The LSP still can't find bits/stdc++.h in a simple file with main, yet it compiles successfully
proofofkeags has quit [Remote host closed the connection]
LevelChart8 has quit [Quit: LevelChart8]
<c00w>
options.foo can be referenced as config.foo
<c00w>
you can also reference boot.initrd.randomoption as config.boot.initrd.randomoption (to read?)
<c00w>
So it feels like there are 4? classes of stuff
<c00w>
option are set in option.foo
<c00w>
defined*
<c00w>
you can set them by referencing foo
<c00w>
or config = { foo=bar }
<c00w>
and then you can read them via config.foo
<gchristensen>
{ config, ... }: { config.bar = config.foo; } an important thing is these are different config's. the one passed as an argument lets you look at the final value of any option
<switchy>
am I doing something odd? I have a hdf5-fortran-mpi package that's something like: pkgs.hdf5.override { inherit gfortran; szip = null; mpi = pkgs.openmpi; } (basically combining the hdf5-fortran and hdf5-mpi packages), but nix-build '<nixpkgs>' -A hdf5-fortran-mpi complains about cyclic references
<switchy>
the error for that being: cycle detected in the references of '/nix/store/3vd443s7xxdyri4h27k9vmizdn59crz9-hdf5-1.10.6-dev' from '/nix/store/vj96g8l8zh65prpckg01wdkqg9366c21-hdf5-1.10.6'
<spinlock[m]>
Does anyone know how to configure NetworkManager to bail quickly if disconnected? I have NixOS running on my laptop, and it's great. 90% of the time, it's hooked up (via usb) to its docked. Through this dock, it has access to the dock's ethernet connection.
<spinlock[m]>
I like that it tries to connect to this wired connection on boot, but (for me) it's not ideal that it blocks user-login
<spinlock[m]>
Unfortunately, if I ever boot this laptop without being connected to the dock, it always waits 90 seconds for the dock's ethernet (dhcp, I imagine?) to timeout.
philr has joined #nixos
<spinlock[m]>
`networking.dhcpcd.wait` looks like it could be relevant, but I'm not experienced enough (yet) to really work out how it functions.
philr_ has quit [Ping timeout: 264 seconds]
hlisp_ has quit [Remote host closed the connection]
drakonis_ has quit [Read error: Connection reset by peer]
<{^_^}>
#88607 (by flokli, 5 days ago, merged): nixos/udev: support /run/current-system/systemd in udevRules
o1lo01ol1o has joined #nixos
<dminuoso>
What's the most elegant way to get this particular fix into my system without switching everything to the nixos unstable channel?
<srhb>
dminuoso: Rebase it on top of whatever commit you want locally and build off that, I'd say. Alternatively you can use disabledModules to disable udev, then import that module specifically from upstream
<srhb>
dminuoso: make sure there's no other changes that make it incompatible with your current commit.
<dminuoso>
disabledModules sounds like a good idea, Ill try that
c0c0 has joined #nixos
opthomasprime has joined #nixos
o1lo01ol1o has quit [Ping timeout: 256 seconds]
opthomasprime has quit [Remote host closed the connection]
proofofkeags has quit [Remote host closed the connection]
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
o1lo01ol1o has joined #nixos
KeiraT has quit [Ping timeout: 240 seconds]
pjt_014 has quit [Ping timeout: 240 seconds]
<magnetophon>
Is anyone using emacs magit for nixpkgs dev? How's the performance?
meh` has joined #nixos
<adisbladis>
magnetophon: Yes, and magit-status is pretty slow
<adisbladis>
Takes a second or two
<adisbladis>
Still worth it
<rnhmjoj>
nixpkgs is just too large. i hope we'll be able to split out nixos and maybe more stuff with flakes
<rnhmjoj>
git status takes 2.8s on my computer
<adisbladis>
I hope we wont
<magnetophon>
adisbladis: Thanks. sure worth it. mine takes much longer though. I have a lot of (old) branches and stashes, does that affect performance a lot?
KeiraT has joined #nixos
<rnhmjoj>
why not?
<lewo>
magnetophon: mine takes between 4 and 5 sec
<tobeportable>
git status 0.04s user 0.09s system 126% cpu 0.105 total
<adisbladis>
rnhmjoj: Having everything in a monorepo is _fantastic_ for various reasons. Testing breakage is much easier for one. Finding stuff is much easier too.
<lewo>
I've seen something to provide a cache to git status. I'm wondering if this could help.
justanotheruser has quit [Ping timeout: 260 seconds]
opthomasprime has quit [Remote host closed the connection]
opthomasprime has joined #nixos
opthomasprime has left #nixos [#nixos]
<rnhmjoj>
lewo: a workaround is to run git status -uno, which avoids checking for untracked files
asymptotically has joined #nixos
<rnhmjoj>
adisbladis: yeah, that's the main pro of a monorepo. unfortunately the performance of git can only get worse over time. also cloning nixpkgs is not fun if you don't have a fast connection
o1lo01ol1o has quit [Remote host closed the connection]
<adisbladis>
I'm well aware, I just don't see the cost of splitting nixpkgs as something worthwhile.
o1lo01ol1o has joined #nixos
<adisbladis>
We could already improve `git status` performance drastically by getting rid of all directories with a single default.nix file
<lewo>
rnhmjoj: not sure it could help with magit: git status takes 0.1s while magit-status takes 4s
mananamenos has joined #nixos
<adisbladis>
Hm, I'm curious how much faster `git status` would be without all those extra dirs
civodul has joined #nixos
<magnetophon>
magit status takes over 20 seconds here. "time git status" takes less than a second for me (seems fishy, something must be wrong with my measurement method,it has definitely taken longer in the past).
noudle has joined #nixos
justanotheruser has joined #nixos
<rnhmjoj>
adisbladis: so, a directory has a fixed cost?
o1lo01ol1o has quit [Remote host closed the connection]
<adisbladis>
magnetophon: File system caches?
o1lo01ol1o has joined #nixos
<adisbladis>
rnhmjoj: Well, every directory has to be read
<evils>
hmm, doesn't nix suffer that cost too?
<adisbladis>
Nix doesn't readdir() every directory
<adisbladis>
We hard code the file path
<evils>
most stuff in all-packages.nix points to a directory, and it then has to find the default.nix?
<adisbladis>
I'm gonna see what difference this actually makes
<rnhmjoj>
aniway, it shouldn't be too hard to test. it's just a matter of renaming the default.nix to the attribute name and update all-packages.nix
<adisbladis>
evils: Hm, possibly it needs to stat() every directory
<adisbladis>
I haven't looked at this in detail
<evils>
while you're at it, rewrite all-packages? :P
<mananamenos>
hi, i have a file which return `pkgs.stdenv.mkDerivation...` expression. How do I install this via configuration.nix systemPackages? I tried (callPackage ./my-file.nix) as well as (import ./my-file.nix). Neither has worked
<adisbladis>
Yeah... We should probably stop using directories to callPackage
chloekek_ has joined #nixos
<magnetophon>
adisbladis: you mean the cache is why "git status" is so fast? Any idea how to troubleshoot why magit is so extremely slow here?
hmpffff has quit [Quit: nchrrrr…]
<adisbladis>
File system caches seems to make this mostly moot (at least for git status)
knupfer has joined #nixos
<adisbladis>
It's around the 200ms mark even after moving blah/default.nix to blah.nix
<adisbladis>
Hm, I need to actually look at what I'm doing here
knupfer has quit [Read error: Connection reset by peer]
<adisbladis>
It seems I inverted some logic :3
cfricke has joined #nixos
<adisbladis>
Holy crap
<adisbladis>
harblasauhteohuceohuc,.äuoä
<adisbladis>
It cut's git status in less than _half_
<adisbladis>
I think this is a worthwhile change
<adisbladis>
magnetophon: rnhmjoj: lewo: ^
<adisbladis>
I have not tried to rewrite all-packages.nix
c0c0 has quit [Quit: WeeChat 2.6]
<adisbladis>
But my assumption is that evaluation times will benefit greatly too
<manveru>
adisbladis++
<{^_^}>
adisbladis's karma got increased to 91
<rnhmjoj>
that's impressive
<{^_^}>
[hydra] @knl opened pull request #771 → Set notificationpendingsince for dependent builds → https://git.io/Jfomy
<rnhmjoj>
adisbladis++
<{^_^}>
adisbladis's karma got increased to 92
<adisbladis>
I only moved files where there was a directory with a single default.nix inside, everything else is intact
<manveru>
maybe a semantic change in the callPackage for top-level only? `if hasSuffix ".nix" then import x else import "${x}/default.nix"` or the like?
<manveru>
would still be an eval cost, but break a lot less history
<adisbladis>
manveru: I thought about that too :)
<manveru>
because if we're gonna restructure nixpkgs, we might as well get rid of those category silliness and use tags instead :)
<adisbladis>
Magit is really noticably happier
<rnhmjoj>
manveru: tags?
<magnetophon>
adisbladis: sorry I read over the thread 3 times, but con't find it. *what* cut's git status in less than _half_?
<adisbladis>
Where there is only one default.nix in a directory
<magnetophon>
adisbladis: ah, thanks
<manveru>
rnhmjoj: atm we have a structure like `pkgs/development/tools/misc/editorconfig-core-c/default.nix`
<manveru>
rnhmjoj: i call those intermediate directories categories
<adisbladis>
manveru: Yeah, those categories don't make much sense to me..
<adisbladis>
I always end up in all-packages to look for the package name anyway
<manveru>
if we had instead `pkgs/editor-config-core-c.nix`, we could even do a `map (i: callPackage i {}) (readDir ./pkgs)` (very simplified) to avoid most of the manual work in all-packages.nix
<rnhmjoj>
yeah, i can never guess the right category at the first try
<adisbladis>
It's also not always clear
<adisbladis>
Is this a development tool, is this a library?
<adisbladis>
Maybe it's both?
<lewo>
I've the same issue with nixos modules: I always look in the module-list.nix file.
<hyper_ch>
anyone runs nixos on a raspi4? I just got message that you can have them now with 8gb... so plenty of ram to use zfs
<evils>
adisbladis++ manveru++
<{^_^}>
adisbladis's karma got increased to 93
<{^_^}>
manveru's karma got increased to 43
<{^_^}>
[hydra] @edolstra pushed commit from @knl to master « Make --no-allow-import-from-derivation configurable in hydra-eval-jobset »: https://git.io/JfoY1
<{^_^}>
[hydra] @edolstra merged pull request #769 → Make --no-allow-import-from-derivation configurable in hydra-eval-jobset → https://git.io/Jfw47
<adisbladis>
manveru: Actually that change doesn't improve anything
<{^_^}>
[nixpkgs] @flokli closed pull request #88158 → Some systemd fixes revolving dbus and socket activation of systemd services → https://git.io/JfuyY
<adisbladis>
The import logic is `lstat(dir)` and if it's a directory then import `default.nix`
<manveru>
ah... right
<adisbladis>
It just reads default.nix directly
<manveru>
it's deeper than callPackage
<adisbladis>
So this change would have to happen in nix itself
<{^_^}>
[hydra] @edolstra merged pull request #771 → Fix: Set notificationpendingsince for dependent builds → https://git.io/Jfomy
<{^_^}>
[nixos-org-configurations] @edolstra pushed to master « Set allow_import_from_derivation to false »: https://git.io/JfoYS
<manveru>
that'd change semantics of nix too much ;|
magnetophon has quit [Read error: Connection reset by peer]
<reirob>
I get: error: The option `programs.ssh.askPassword' has conflicting definitions, in `...'.
<reirob>
Sorry, let me copy the whole error, it might be important:
<reirob>
error: The option `programs.ssh.askPassword' has conflicting definitions, in `/nix/var/nix/profiles/per-user/root/channels/nixos/nixos/modules/services/x11/desktop-managers/plasma5.nix' and `/nix/var/nix/profiles/per-user/root/channels/nixos/nixos/modules/programs/seahorse.nix'.
<manveru>
hmm
<reirob>
I am quite new to nixos
<reirob>
So I am looking for advice what to do
<reirob>
I believe it is because in 19.09 I had a lot of desktop environments installed, especially gnome and kde
<manveru>
`programs.ssh.askPassword = mkDefault "${plasma5.ksshaskpass.out}/bin/ksshaskpass";` vs `programs.ssh.askPassword = mkDefault "${pkgs.gnome3.seahorse}/libexec/seahorse/ssh-askpass";`
<reirob>
But it worked fine. Now on 20.03 I interpret it that there is a conflict between kde plasma and seahorse
<manveru>
so, you gotta choose one
Neo-- has joined #nixos
<manveru>
by putting something like `programs.ssh.askPassword = "${pkgs.gnome3.seahorse}/libexec/seahorse/ssh-askpass";` in your configuration.nix
<{^_^}>
[nix] @edolstra merged pull request #3565 → Use nixpkgs#bashInteractive for dev-shell → https://git.io/JfZvq
<reirob>
@manveru, many thanks, I will immediately try it out!
<reirob>
And will tell if it worked.
<manveru>
i think that's the entire point of the searhorse module, i'm surprised they set it to `mkDefault`
<reirob>
I am wondering how it worked in 19.09
justanotheruser has quit [Ping timeout: 272 seconds]
<manveru>
the seahorse one was only introduced in october last year
<manveru>
so that explains that :)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<reirob>
Ah, that explains it! Thank you. I can confirm, that by adding programs.ssh.askPassword = "${pkgs.gnome3.seahorse}/libexec/seahorse/ssh-askpass"; now the nixos-rebuild seems to work, it's still working, but I can see that it continues further than where it was
<reirob>
Many thanks @manveru :)
<reirob>
I am wondering, how did you find out what the problem was, so quickly?
drakonis1 has quit [Read error: Connection reset by peer]
drakonis1 has joined #nixos
<nixbitcoin>
Hey, I'm still working on creating a generalized netns creator service. This is the first function I've ever written, and I'm struggling with syntax and a lib.forEach. You can see my most current mkNetns function here: https://gist.github.com/nixbitcoin/72c5b5b77471775f5d28eb98ec7ba22b
<nixbitcoin>
Specifically I can't seem to figure out where to define the function so I can call it from other modules and how to combine variables from the function's argument set and other files.
vidbina has joined #nixos
<nixbitcoin>
Can somebody point me to similar code in the nixpkgs or help me with this function specifically? I would be very grateful.
cosimone has quit [Quit: Quit.]
<{^_^}>
[nixpkgs] @alexbiehl opened pull request #89099 → python3Packages.skein: init at 0.8.0 → https://git.io/JfosS
<kloenk>
Im a little lost why this build fails on my x86_64-linux hydra. Can somebody point me in some directions? https://hydra.kloenk.de/build/262
<b42>
nixbitcoin: since it's touching `config` i suppose the easiest way to call it from other modules would be to expose it as a module itself and "call" it by setting its options?
<nixbitcoin>
b42: I want to have a different "netns-" service for every nix-bitcoin service that binds to its nix-bitcoin service and creates its netns before nix-bitcoin service starts. For this I wanted to create a function that is imported in every module with something like `systemd.services.netns-bitcoind = mkNetns { name = "bitcoind"; iface = "veth1"; };`
<nixbitcoin>
b42: If it were a standalone module, could I set it's options multiple times for every service?
drakonis1 has quit [Ping timeout: 272 seconds]
<b42>
nixbitcoin: you can - some types of options can be merged so if you e.g. declare an option "services.netns" with type = types.attrOf types.str then one module can set services.netns = { bitcoind = "veth1"; }; other module can set services.netns = { clightning = "veth2"; }; and your module will see it as { bitcoind = "veth1"; clightning = "veth2" }
<nixbitcoin>
How would I then parse the set for the correct name's and veth's for every netns creator service?
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<b42>
in the conifg = ... part of your module you'd do something like systemd.services = mapAttrs (name: iface: { /* ...what you have now in mkNetns */ })
<b42>
nixbitcoin: ^ if i understand the question correctly
<b42>
so systemd.services = mapAttrs' (name: iface: nameValuePair "netns-${name}" { /* service definition */ })
ZaraChimera has quit [Quit: Going offline, see ya! (www.adiirc.com)]
ok2` has quit [Ping timeout: 240 seconds]
ok2`- is now known as ok2`
<nixbitcoin>
Ah cool, I'll give that a try right now. Do you have any ideas how to solve the whole availableNetns thing where I'm trying to add lines to the ExecStart list for every netns in the service's availableNetns list?
<__monty__>
Hmm, so sqlite is already a multiple-output expression and the default output *is* bin. It's just that the binary is built without readline support by default.
alp has quit [Ping timeout: 246 seconds]
<b42>
nixbitcoin: don't know the context here - does every netns need to have list of other namespaces it's allowed to talk to?
<__monty__>
1400 occurrences of sqlite without an output specified. Would changing the default to interactive and specifying .dev and .out everywhere the binary isn't needed even be accepted in a PR?
calbrecht has quit [Remote host closed the connection]
Darkmatter66 has quit [Ping timeout: 260 seconds]
Darkmatter66_ has joined #nixos
kvda has joined #nixos
gilligan has quit [Quit: WeeChat 2.7.1]
Darkmatter66_ has quit [Quit: ZNC 1.7.5 - https://znc.in]
Chiliparrot has joined #nixos
piezoid has joined #nixos
<manveru>
,locate bin rpmbuild
<{^_^}>
Found in packages: python27Packages.rpm, python37Packages.rpm, python38Packages.rpm
<b42>
to account for the "enable" option your interface would have to be something like { enable = mkEnableOption "..."; namespaces = mkOption { type = types.attrsOf (types.submodule ...)); } }
<{^_^}>
[nix] @edolstra merged pull request #3631 → Add unit tests for config.cc → https://git.io/JfrrW
<codygman___>
Is there a tool to figure out what change is causing nix-build to rebuild? I have a repo with a Haskell project and a README.md. I ran it once and pushed to cachx and I see it's pushed there successfully. I see my custom dependencies are pulled from cachix. Then I modify the README.md and it causes the whole Haskell package to be rebuilt even
<betawaffle>
clever: does the order of modules passed to evalModules matter? for example, are those args only available to modules later in the list
hmpffff_ has joined #nixos
<infinisil>
betawaffle: They're available to all modules, order doesn't matter
<betawaffle>
fancy
orivej_ has quit [Ping timeout: 246 seconds]
<infinisil>
(unfortunately module order sometimes does matter for other things, ideally it shouldn't)
hmpffff has quit [Ping timeout: 260 seconds]
orivej has joined #nixos
<betawaffle>
yeah, i'm not worried about that. ok, next question... does a "configured" nixpkgs (ie. `import <nixpkgs> {}`) include a reference to itself somewhere?
<betawaffle>
for example, is there an argument i can add to a package to get access to the nixpkgs directory, like for referencing nixos modules
NeoCron has quit [Ping timeout: 260 seconds]
<clever>
> pkgs.path
<{^_^}>
/var/lib/nixbot/nixpkgs/master/repo
<betawaffle>
perfect!
magnetophon has quit [Read error: Connection reset by peer]
<betawaffle>
maybe final question: are there any default arguments available to modules given to evalModules? ie. `(lib.evalModules { modules = [ (args: { config.moduleArgs = args }) ]; }).config.moduleArgs`
chloekek_ has joined #nixos
jlv_ has quit [Ping timeout: 256 seconds]
<{^_^}>
[rfcs] @lheckemann closed pull request #64 → [RFC 0064] New Documentation Format for nixpkgs and NixOS → https://git.io/JejtS
hmpffff_ has quit [Quit: nchrrrr…]
<betawaffle>
looks like config, lib, and options
jlv_ has joined #nixos
zaeph has quit [Ping timeout: 256 seconds]
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
<{^_^}>
[nixpkgs] @rnhmjoj pushed 2 commits to release-20.03: https://git.io/JfolB
magnetophon has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @NeQuissimus pushed 6 commits to release-20.03: https://git.io/Jfolp
Jeroen[m]1 has left #nixos ["User left"]
wadkar has joined #nixos
<wadkar>
hi, fresh install of Catalina with `xcode-select --install` done; but installer complains 'you do not have xz installed', what gives?
<wadkar>
(I obviously don't wanna install brew/macports for this, as I want to stay within nixpkg boundaries)
<aveltras>
any idea why i'd have to recompile so many things when using a shell.nix like the following ? I'd think pandoc is a common enough library that it would be downloaded from the cache
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej_ has joined #nixos
<CRTified[m]>
and no application can record anything with it. I can also only choose ALSA for audio interfacing, although I use pulse. There is also no pulse entry, but just a large list of actual audio hardware available in my system. I'm on nixos-20.03.1950.48723f48ab9 as my main channel right now.
<CRTified[m]>
(I was redirected from #nixos-chat to maybe ask here): Hi, I have a problem with audacity and pulse, maybe someone here has an idea. When starting, I get this output: https://gist.github.com/CRTified/d424e8c07a1dd8590b1a69855250d549 - and from there on, I can't use the microphone input of my onboard sound card. Looking at the VU-Meter in pavucontrol suggests that there is just ~1 sample/second coming from the soundcard,
LevelChart8 has joined #nixos
magnetophon has joined #nixos
<wadkar>
I can't seem to find the manual steps required to do `--darwin-use-unencrypted-nix-store-volume` step. The install script provides a URL with anchor that doesn't exist: https://nixos.org/nix/manual/#sect-macos-installation
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
j-k has joined #nixos
jlv_ has quit [Quit: leaving]
orivej_ has quit [Ping timeout: 256 seconds]
orivej has joined #nixos
<LnL>
yeah the manual hasn't updated yet for some reason
<infinisil>
Not sure about the steering committe meeting, but for the RFC meetings I wouldn't mind if more people asked to observe
<evils>
emphasis on asking or observing?
codygman___ has quit [Remote host closed the connection]
simba2 has quit [Ping timeout: 272 seconds]
ddellacosta has joined #nixos
alexherbo2 has joined #nixos
gustavderdrache has joined #nixos
<niksnut>
we also just wondered if they should be recorded
acowley has quit [Ping timeout: 240 seconds]
<evils>
if at all possible, that seems like a great option
acowley has joined #nixos
simba2 has joined #nixos
turion has quit [Quit: Leaving.]
<j-k>
Looks like a meeting just finished on flakes, comment posted on the issue 2 mins ago moving it to FCP
xourt has joined #nixos
<bqv>
?!
<gchristensen>
wooo!
<bqv>
awesome
<gchristensen>
niksnut: we could put record them and put them up on the youtube channel. that said, I think some people don't want to be recorded -- so maybe not all of them
fendor_ has joined #nixos
<j-k>
the sprunge link for the meeting notes isn't working over https and by browser keeps redirecting there ... curl -> pager to the rescue
<j-k>
*my browser
fendor has quit [Ping timeout: 260 seconds]
<bqv>
so what does the rfc being merged even change? presumably channels aren't being deprecated, and i've heard a lot of cries of "it isn't even merged yet" but there isn't really any breaking change
<bqv>
is this essentially just a statement of intent
Chiliparrot has joined #nixos
<niksnut>
acceptance of an rfc means we're agreed in principle on some feature, but not necessarily the implementation or every detail
hmpffff has quit [Quit: nchrrrr…]
reirob has joined #nixos
<reirob>
Hello again #nixos :)
<bqv>
gotcha
<reirob>
I have a new problem. I installed a program called pokerth. When I log in to KDE then it works, but when I try to start it in Gnome it crashes
<reirob>
The error that it gives: qt.qpa.plugin: Could not find the Qt platform plugin "xcb" in ""
<reirob>
This application failed to start because no Qt platform plugin could be initialized. Reinstalling the application may fix this problem.
orivej has quit [Quit: No Ping reply in 180 seconds.]
<reirob>
Available platform plugins are: wayland-org.kde.kwin.qpa.
<reirob>
Aborted (core dumped)
<{^_^}>
[nixpkgs] @eadwu opened pull request #89112 → musl: include libc.musl-$arch → https://git.io/JfoRv
<reirob>
Any hints what it might be?
<reirob>
Or what I can do about it?
orivej has joined #nixos
<reirob>
I am happy to submit a bug to nixos. But I never did, so I would need some start help, about how to do this
<reirob>
If that's the way to go
<j-k>
the nixpkgs repo has an issue template that's very helpful if you'd like to fill one out. And as usual, provide as much detail as possible
<alexarice[m]>
reirob: This is quite a common problem with qt programs and should be easish to fix
<reirob>
@alexarice, what should I do?
<reirob>
I guess if it is a problem with qt programs, then I am not the only one for whom it happens.
<j-k>
reirob: yup that repo. & yeah it's a QT issue, works in KDE because it's QT based. Not sure what the package is missing as I haven't done any QT packaging yet
<alexarice[m]>
reirob: I had a look at the derivation, it looks like it has been fixed, are you on 20.03?
<alexarice[m]>
*has been fixed on unstable
<reirob>
yes I am on 20.03
<j-k>
updated your channel?
<j-k>
looks like it hasn't made it's way to the nixos-20.03 branch yet
<alexarice[m]>
either you could make an issue asking for a backport and cc nixos-backports (?) or get the version from unstable
eoli3n_ has quit [Ping timeout: 264 seconds]
<reirob>
ah, but it will probably land in 20.03 and then when I will do an upgrade it will work, right?
<reirob>
I guess I will just open an issue and provide my details, right?
<alexarice[m]>
reirob: it probably won't land in 20.03 if no one knows it is broken
<alexarice[m]>
i.e. yes you should make an issue
<reirob>
thank you very much for your nice and helpful advice, people. I will try now to open an issue
<{^_^}>
#89113 (by reirob, 20 seconds ago, open): pokerth in gnome3 crashes, but works in plasma5
<reirob>
:-)
<alexarice[m]>
reirob: I have ccd them in my message on the pr so don't worry about it
f0x has quit [Quit: killed]
joepie91 has quit [Quit: killed]
<reirob>
Thank you very much for your help.
cantstanya has joined #nixos
jlv has quit [Disconnected by services]
f0x has joined #nixos
jlv has joined #nixos
<reirob>
Now I will try to install from the unstable :)
<jlv>
Is there something like overlays for modules? In particular, there is a pull request I want to test.
Guest30241 has joined #nixos
<bqv>
i had a system for something like that, once
<bqv>
but generally, no
<bqv>
there's an option for disabling load of certain modules, and you can fetch the PR and import the module manually from there, i guess. but generally yeah, it's a pain in the ass
<jlv>
bqv: at least that sounds better than maintaining my own fork.
<bqv>
well, my other system was to automatically fetch several PRs and patch them onto the current version of nixpkgs
<bqv>
which is even nicer, but is also basically a crappy solution to a crappy situation
<jlv>
bqv: does it make any difference if I want to patch a home-manager module? I already fetch the home manager git repository before importing it, so I guess I could patch it before the import.
<bqv>
yes, it makes all the difference
<bqv>
just patch it manually. anything else will be messier
fendor_ has quit [Remote host closed the connection]
<aveltras>
is there any way to know what is in progress when entering a nix-shell ? i've been staring at my terminal for more than 20min now, i know it's downloading something but i don't know what
<alexherbo2>
bqv even with the examples I do not understand
<clever>
bqv: try without NIX_REMOTE, that might be what broke it
<hyperfekt>
aveltras: try nix-top
<bqv>
no, that's what fixed it
<alexherbo2>
clever: thanks
<bqv>
i had to do that to disable the flake-y daemon
<bqv>
otherwise there's no point to using nixpkgs#nix
<clever>
bqv: you want NIX_REMOTE=local to disable it
<bqv>
oh, ok
<bqv>
well, either way it worked
<bqv>
now i want to know what it is about new nix that breaks it
urkk has joined #nixos
<aveltras>
hyperfekt: thanks, i try that
<bqv>
presumably it's something to do with writeReferencesToFile
<urkk>
Hi! I'm trying to install nix withing Archlinux and the `nix-channel --update` command fails with `executing '/usr/bin/bash': No such file or directory`
<urkk>
It looks like the chroot lacks /usr/bin/bash, so its complaining
<alexherbo2>
bqv, clever: you are life saver, thanks
<bqv>
np
<alexherbo2>
I do not understand what I was doing though
<reirob>
In the meantime I managed to install the unstable
<alexarice[m]>
probably a day or 2
asbachb has quit [Ping timeout: 245 seconds]
<reirob>
And it works, though for the hidpi issue I am using the workaround that I put in the notes
<reirob>
Cool, now I know how to submit bug tickets :D
justanotheruser has joined #nixos
<bqv>
alexherbo2: that second file makes no sense to me
<reirob>
I guess I will come here more often, until I know how to use nix and nixos good enough :)
<bqv>
explain what you're trying to achieve, might help
<reirob>
Have a good time all :) Checking out
reirob has quit [Client Quit]
philr has quit [Ping timeout: 260 seconds]
lord| has joined #nixos
<evils>
i'm still stuck at `ModuleNotFoundError: No module named 'setuptools'` in buildPythonPackage on trying to build wxPython 4.1.0; help would be welcome
wfranzini1 has quit [Remote host closed the connection]
<urkk>
This is the child full command: /usr/bin/nix-env --profile /nix/var/nix/profiles/per-user/ram/channels --file <nix/unpack-channel.nix> --install --from-expression f: f { name = "nixpkgs-20.09pre226586.571212eb839"; channelName = "nixpkgs"; src = builtins.storePath "/nix/store/3ivzcbfdw9hn9vcf9lp2b6vkjm1al4bk-nixexprs.tar.xz"; } --quiet
<LnL>
hmm, is there a way to add an indirect root for just a path (like a drv)
<clever>
LnL: i typically use `nix-store -r $foo --add-root result --indirect`, but the `-r` will try to build a drv first
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<LnL>
does that work if it's built?
<LnL>
would kind of expect that to link the output
<clever>
LnL: it would be a result pointing to the $out, not the .drv file
<LnL>
yeah so I have a drv that I copied over and want to keep all the build deps alive for
<clever>
not sure how to do that
<clever>
i tend to use a local hydra for that job
<LnL>
I'll just create a root, but then I'll probably forget about it
j-k has joined #nixos
mthst has quit [Ping timeout: 265 seconds]
KindTwo has joined #nixos
cosimone has quit [Quit: Quit.]
KindOne has quit [Ping timeout: 272 seconds]
<urkk>
nvm, it looks I needed to setup a sandbox with some coreutils first with bash
KindTwo is now known as KindOne
orivej has quit [Read error: Connection reset by peer]
orivej_ has joined #nixos
ryantm has quit [Ping timeout: 265 seconds]
drakonis1 has joined #nixos
<evanjs>
Bah. Anybody know of any open ovf tools for linux or maybe even on NixOS? Running into this when trying to import an OVA (virtualbox) generated nixos-generators
<cole-h>
betawaffle: Because of the `modSha256 = null` I'd bet.
<betawaffle>
no matter what i try for subPackages, i get some form of that error
<cole-h>
,tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected. See: tofu-vim
<betawaffle>
is there anything i need to change to upgrade to `vendorSha256`? modSha256 is complaining loudly about being deprecated, but when i try to switch, it seems like maybe there are some build errors?
<betawaffle>
nevermind
evanjs has joined #nixos
<energizer>
is it right to think of `or` and `.` as binary operators, with `or` having higher precedence than `.`?
<ldlework>
What's a function that will return an attrset without the given attrs?
<craftyguy>
if I make a derivation of, e.g. firefox, to use a newer version than what's in the channel I am subscribed to and later on the channel gets an even newer version than what I have in my derivation, which takes precedence on a 'switch --upgrade'? is there a way to have nix always prefer the more recent version of the package (whether in user nix profile or in the subscribed channel)?
<ldlework>
craftyguy: you can write some kind of helper function to compare the version in nixpkgs to your override and pick the one with the highest version
<ldlework>
then use your helper everywhere
zupo_ has joined #nixos
<ldlework>
no idea if this is the best way, but it is a way
<cole-h>
If your derivation of firefox is an overlay, it will take precedence no matter what (because the firefox from nixpkgs no longer exists due to you overlaying it with your firefox)
<craftyguy>
so in my example it might be 'firefox-latest' and it looks in nixpkgs and my override to find the most recent
zupo has quit [Ping timeout: 246 seconds]
<ldlework>
If you use an overlay you can't see both versions at the same time
<craftyguy>
cole-h: ok, I guess I don't understand what an overlay is. time for more reading!
<ldlework>
You'd want to use an overrideAttrs conditionally
* craftyguy
new here, and trying to figure all this out
<simpson>
ldlework: It's legible. I don't know enough about Styx to know what the Styx style should look like, but I can guess at how to extend this module by imitating it.
<ldlework>
craftyguy: an overlay basically changes what is in nixpkgs
<ldlework>
simpson: nice thank you for the reply
<ldlework>
simpson: the aim is for styx to be super simple. it will flatten your outputs, and evaluate their renderer function then write the result to their path attribute
<simpson>
No worries. Some code review is easier than others.
<ldlework>
so that it is apparent how one would extend it is good, because the idea is there shouldn't really be a baked-in style. it should be programmable.
<simpson>
Sure, makes sense.
<ldlework>
<3
<simpson>
Tangent: Is there a good GH query for PRs to nixpkgs which need reviews from non-owners?
<citizenzibb[m]>
I would like to run nixos in a virtualbox image. what are the optimal settings? or can I just go with whatever virtualbox provides?
<simpson>
Finding myself with more and more spare time, and actually starting to reply to long-dead GH issues to get my queues cleaned up. I updated the PR I had forgotten about, but I want to know what else needs help.
<simpson>
energizer: Well, TBF, the workflow works fine for folks who are part of it. If I'm a maintainer, which is rare but possible, then my review *does* mean something and is part of the process.
<citizenzibb[m]>
simpson: I suppose I will just follow that
<Cadey>
is there a way to set the default nice level of nix builds so they are slightly lower priority than everything else?
<simpson>
But I'm noticing that, more and more, some sort of informal thumbs-up is also preferred/valued somehow, and I guess I'm wondering about that.
<cole-h>
energizer: Because we're at 2k+ PRs. Nobody is going to wade through all of them. If you want your PR reviewed, you gotta put in some more effort (e.g. posting in that thread)
<energizer>
cole-h: not sure i understand. isn't the goal to make less work rather than more
<cole-h>
energizer: Not sure I understand, either. There are so many PRs that things will inevitably fall through the cracks. Meaning it's nice to have some other way to get the attention of people who can merge.
<cole-h>
-> meaning the GH PR flow is not enough. Hence the presence of that thread
<euank>
specifically the overlay having something like `firefox = if super.firefox.version > my-firefox.version then super.firefox else firefox;`
<euank>
since you have a reference to super, you can do stuff based on its version
<craftyguy>
euank: that's a really helpful example, thank you!
<euank>
There's probably a better version comparison function in nix, but I don't know offhand
<simpson>
Right. But what exactly is the bottleneck? Do we need more ofborg? More automatic LGTM or thumbs-up on ryantm PRs? Better automatic updates for language-specific subsystems?
<craftyguy>
np. it at least gives me some idea of how this would work
<euank>
Good luck
zupo_ has quit [Ping timeout: 260 seconds]
tureba has joined #nixos
quinn has joined #nixos
<hyperfekt>
okay so. nix complains about a hash mismatch and reports two identical hashes for wanted and got. wtf?
<cole-h>
simpson: r-ryantm PRs get merged with relative frequency by e.g. jonringer and friends. What really helps is two things, IMO: 1) having interested parties review the contents and functionality of the PR; and 2) have interested parties merge those PRs.
hmpffff has quit [Quit: nchrrrr…]
magnetophon has joined #nixos
<Cadey>
is there a way to set the default nice level of nix builds so they are slightly lower priority than everything else?
<cole-h>
I don't think that just shotgunning "LGTM"s are really helpful, unless you also test functionality to the best of your ability.
<energizer>
Cadey: the goal being to ensure they get killed first?
<simpson>
cole-h: Hm. (1) means more maintainer attention to PRs. (2) means more people with merge abilities.
<cole-h>
Indeed.
<Cadey>
energizer: the goal being to ensure they are lower priority than my firefox instance
magnetophon has quit [Read error: Connection reset by peer]
<cole-h>
Every PR I get review-requested for gets a review (if I'm the maintainer) because of this. Then, it's not up to me anymore -- somebody needs to merge it, and that someone isn't me.
zupo has joined #nixos
<energizer>
Cadey: i mean what do you want that priority to affect?
<adisbladis>
2) may also mean tooling assisted merges from people with less privileges overall
<Cadey>
energizer: i want to make sure that nix builds and their related processes don't make my firefox session lag
<cole-h>
That too, but I think that needs an RFC (and the related one got closed due to inactivity IIRC).
<adisbladis>
Yeah, there's a lot of issues to solve around increasing the bandwidth
<energizer>
Cadey: i wonder if you could wrap `nix-build` with `nice`?
<adisbladis>
I'm not sure just adding more people with merge ability will really solve that
<simpson>
Cadey: I can't find anything offhand in nixpkgs or NixOS for this. I might start by examining how the NixOS service configures the Nix daemon, and looking at whatever systemd offers for nice'ing.
<adisbladis>
And even if it does what's the associated cost?
<craftyguy>
nbathum: that makes it seem like there is a built-in thing to return the latest version given a list of derivations?
<simpson>
adisbladis: I usually use https://xkcd.com/1205/ for this. I estimate that we could justify a few months per language subsystem just for the topic of automatic merges and defining automatic testability.
<simpson>
Also adisbladis++ for systemd incantations.
<{^_^}>
adisbladis's karma got increased to 94
<{^_^}>
[nixpkgs] @FRidh pushed 231 commits to staging-next: https://git.io/JfoK8
<adisbladis>
Oh, sorry
<energizer>
adisbladis: does that apply to children?
<euank>
so not builtin to nixpkgs upstream, but it's an easy enough one-line function to copy and looks like exactly what you want
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<craftyguy>
yeah, that's great. thank you
<nbathum>
thanks euank++ for explaining
<{^_^}>
euank's karma got increased to 4
virus_dave has quit [Ping timeout: 265 seconds]
virus_dave_ is now known as virus_dave
<adisbladis>
energizer: It's inherited (I'm like 95% sure)
never_released_ has quit [Remote host closed the connection]
<craftyguy>
what's this karma thing here?
<adisbladis>
I mean, the nice level is inherited by the parent process on linux normally
<adisbladis>
I don't think nix-daemon is doing anything special here
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @davidak opened pull request #89124 → geekbench5: add compute benchmark support → https://git.io/JfoKM
<energizer>
adisbladis: what's the next step toward migrating nixpkgs to using poetry2nix?
alexherbo2 has quit [Ping timeout: 256 seconds]
<adisbladis>
energizer: Just start doing it?
<adisbladis>
Go package by package
<energizer>
adisbladis: only for poetry packages, or is there a way to do it for setuptools packages too?
<adisbladis>
energizer: You can install setuptools packages just fine with poetry2nix
<adisbladis>
In fact, the initial use case was packaging something that wasn't using poetry
<energizer>
adisbladis: got a link?
magnetophon has joined #nixos
never_released has joined #nixos
<chreekat[m]>
I do wish I could set niceness on builds independently. It would have to be an argument to nix-build sent to the daemon as part of the job spec and applied explicitly to the builder process by the daemon,I guess, but I'm just waving my hands here.
hmpffff has joined #nixos
<euank>
cgroups are a better way to set cpu priority and optionally limit cpu usage of stuff
<{^_^}>
nix#159 (by vlstill, 6 years ago, open): Cgroup
<euank>
Which would have used the caller's cgroup. That seems like a quite elegant way to handle this
cjpbirkbeck has quit [Remote host closed the connection]
<euank>
i.e. you could run `systemd-run --scope --property=CPUWeight=100 -- nix-build ...` I think, and if the daemon ran builds in your same cgroup, it'd get that cpu limit
<euank>
.. but it looks entirely stalled out at this point, so it doesn't actually help
seanpars- has joined #nixos
<energizer>
i've been trying to use cgroups v2 to limit ram but havent been successful on nixos
<bqv>
euank: cgroups were never intended for use in the way systemd uses them, i could understand hesitance on depending on that and even on depending on systemd more than currently necessary
cr4y1 has quit [Remote host closed the connection]
<energizer>
argparse was added to stdlib in 3.2
cr4y1 has joined #nixos
<adisbladis>
energizer: I can swear it was in 2.7 already?
<energizer>
adisbladis: backport
<evanjs>
urgh
<evanjs>
I don't wanna deal with optionals but blah
<evanjs>
I'll try both I guess
<evanjs>
pythonOlder and etc
<energizer>
evanjs: it's not optional, all maintained versions of python have it in stdlib
<adisbladis>
evanjs: Don't bother
Ilya_G has quit [Remote host closed the connection]
chloekek_ has quit [Ping timeout: 258 seconds]
lorimer has quit [Ping timeout: 265 seconds]
<betawaffle>
does buildGoModule create derivations for each dependency, or does it clump all the dependencies into one blob?
<energizer>
oh i have my timeline backwards, 2.7 was released a few months before 3.2
<betawaffle>
it would be cool if it could share already-built deps between different modules, if they are the same version
<adisbladis>
Blobby McBlobface
lorimer has joined #nixos
<betawaffle>
is that an answer to my question adisbladis?
<evanjs>
oh okay sweet thanks guys
<adisbladis>
betawaffle: Yes ^_^
<betawaffle>
adisbladis++
<{^_^}>
adisbladis's karma got increased to 95
<evanjs>
what about "psutil>=5.6.7; python_version < "3.3"',"?
<energizer>
evanjs: not stdlib
<evanjs>
also, for those that were linked to my version functions earlier (craftyguy?) the python version functions might be helpful as well
<energizer>
maybe they're using some function that was added to another stdlib module
<energizer>
anyway 3.3 is very eol
<evanjs>
And versionAtLeast and the related base functions, maybe
<evanjs>
energizer: yeah no kidding
<evanjs>
adisbladis: oh lol that initial package I linked had python2Packages imported. Should've put 2 and 2 together there (re argparse) since they were indiscriminately prePatching lol
<betawaffle>
and the readme talks about needing to use a tool called modvendor, if you want to use vendoring
<betawaffle>
who maintains the buildGoModule functionality?
<betawaffle>
kalbasit: you, perhaps?
<adisbladis>
betawaffle: kalbasit (and I think Mic92 also has something to do with it)
<adisbladis>
evanjs: Eww, cot is doing some pretty disgusting things
remirol has joined #nixos
<kalbasit>
betawaffle: I started it but it's maintained by a number of folks now. c00w in particular
<evanjs>
adisbladis: yeah.... that's what I'm dealing with rn...
<kalbasit>
betawaffle: do you have a particural question for me (sorry in a work meeting, have not followed the chat)
<evanjs>
Trying to find _something_ that helps me deal with OVF files (vs vmware crap) and I stumbled across this
<evanjs>
right now, I'm wondering why this test is failing, and if it matters that cot itself isn't in the path for the tests...
fendor has quit [Read error: Connection reset by peer]
<evanjs>
that requirements.txt, tho...
<betawaffle>
wanted to know if you know much about the switch from modSha256 to vendorSha256, because it breaks one of the builds i'm trying to do
<evanjs>
"I need myself" lol
<adisbladis>
evanjs: i looked at install_helpers.py
lorimer has quit [Ping timeout: 246 seconds]
Chiliparrot has joined #nixos
<evanjs>
adisbladis: u-uh-oh.....
remirol is now known as lorimer
<adisbladis>
evanjs: energizer: I saw this as an opportunity to show energizer what it looks like using poetry2nix to package non-poetry apps http://ix.io/2nE8/nix
<kalbasit>
betawaffle: that was done because modSha256 is the sha256 of the `$GOPATH/pkg` folder (or at least part of it) and the Go team has already made two changes there breaking all of our shas. The `vendor` directory however is fixed and is supported fully by Go as opposed to what we initially did which was a hack-ish
<energizer>
adisbladis: thanks
<kalbasit>
betawaffle: shouldn't be too hard to fix though, just remove modSha256 and put vendorSha256 and take in the new sha. should be good to go
<betawaffle>
that makes sense
<betawaffle>
kalbasit: what i mean is that the vendoring process doesn't copy all the necessary files for the build
<kalbasit>
betawaffle: huh ok that sounds like a bug. Mind filing an with steps to reproduce? I can take a look after work
<betawaffle>
sure
<{^_^}>
[nixpkgs] @costrouc opened pull request #89127 → python3Packages.diagrams: init at 0.12.0 → https://git.io/Jfoi6
<kalbasit>
betawaffle: thx! Will take a look. Would be great to have an issue though so we can discuss it there!
<betawaffle>
sure, i'll make one when i get off work
<tobiasBora>
Hello, I'd like to do the following: create a program that always compiles for a specific arm architecture (I use the cross compilation chain arm-linux-gnueabi-gcc)
<tobiasBora>
However I'm not sure to see how to specify that I want as input a library (in my case libdirectfb) compiled for arm
<zeta_0>
are any of you running into this error when running `lorri shell` ? `/nix/store/b4kbdvdyb80r57jhx7665117k75v4jrl-stdenv-linux/setup: line 795: /build/env-vars: No such file or directory
<infinisil>
From this you can see that you can pass modBuildPhase to override the default build phase, modInstallPhase to override the default install phase, or you can pass preBuild/postBuild/preInstall/postInstall to have commands run before/after the default build/install phase
<eyJhb>
What is the difference between `${lib.getBin pkgs.xorg.xrandr}/bin/xrandr --setprovideroutputsource 2 0` and `${pkgs.xorg.xrandr}/bin/xrandr --setprovideroutputsource 2 0`
<eyJhb>
Why would one use lib.getBin
<infinisil>
aveltras: And wherever you want to, you should be able to call `make prod` directly
<hr[m]>
If possible, could a nixpkgs maintainer with merge rights take a look at PR #88922 The assigned reviewer has not ben active in a few months and has not merged previous PRs.
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<evanjs>
stoooop using git commands in the Makefile lol
lassulus_ has joined #nixos
<bqv>
yikes
<bqv>
(that's a terrifying makefile anyway, jeez)
<evanjs>
yeah.....
<evanjs>
"Linux: Just run `make`"
<evanjs>
Uh huh...
lassulus has quit [Ping timeout: 256 seconds]
lassulus_ is now known as lassulus
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
noudle has quit []
lsix has quit [Quit: WeeChat 2.8]
<hexagoxel>
any downsides to `importOrElse = maybePath: otherwise: if builtins.pathExists maybePath then import maybePath else otherwise`, and does it exist already?
<{^_^}>
[nixpkgs] @Mic92 merged pull request #86055 → nixos/wireguard: test against multiple kernel versions → https://git.io/JftSW
<clever>
hexagoxel: i have basically the identical code in here
<evanjs>
And some of the deps use meson.... what
<euank>
evanjs: I had one package where I ran into that problem and considered solving it by `let git = writeShellScriptBin '' case $1 in log) echo ... exit 0 .... etc etc ''
zeta_0 has joined #nixos
<euank>
And added that to the path in the builder via buildInputs. Which let me have a bash script that printed a fixed revision (the one I was fetching via src anyway) for git describe and git rev-parse etc
<hexagoxel>
clever: interesting
<euank>
I thought that it might make sense to actually go back and implement that idea as a proper "mockgit" program in rust or some real language that actually can parse arbitrary git commands, and doesn't depend on the `.git` directory existing
<clever>
hexagoxel: CI will lack secrets, and only test if the code compiles, but locally, secrets.nix does exist
<euank>
but instead has some more reasonable input the nix packager could pass to it to describe "here's your commit and tag" etc
<hexagoxel>
although I have "local/dev-specific nix config" in mind rather than secrets
<euank>
evanjs: Does that sound like something that would be helpful? I guess for that makefile, it's a mess of submodules, which my hack wouldn't help with
<hexagoxel>
things like "test against the updated version of this dependencies that I cloned into this other folder"
zeta_0 has left #nixos [#nixos]
<euank>
I wanted that hack specifically so I could avoid needing to clone .git
<evanjs>
euank: yeah.... that's what I just realized lol. it needs .git for the crap they're doing so this is a bit more than I figured at first
<betawaffle>
Mic92: I'm trying to use modvendor right now as a workaround, and it doesn't seem to be picking up my postBuild
<betawaffle>
is there some trick?
<Mic92>
betawaffle: are you using overrideModAttrs ?
<aveltras>
does anyone have an example or building a node package involving a gulpfile ?
<eyJhb>
Will persist that then!
<eyJhb>
:D
xwza91 has joined #nixos
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<xwza91>
Hey, I would be glad if someone could help me - I've (seems to be successfully) upgraded from nixos 19 to 20.03 - if I do nixos-rebuild switch I get following warning: "warning: In file /nix/var/nix/profiles/per-user/root/channels/nixos/nixos/modules/tasks/encrypted-devices.nixa list is being assigned to the option config.boot.initrd.luks.devices."
<xwza91>
my only config to this option is "initrd.luks.devices.bootsec.device = "/dev/disk/by-uuid/<UUID>";" so my gues is the config in /nix/var/profiles is outdated? Is there already a fix? how can I install it?
<xwza91>
clever: Same "devices = map (dev: { name = dev.encrypted.label; device = dev.encrypted.blkDev; } ) keylessEncDevs;" but this file is not created/maintained by myself - so how to update if "nixos-rebuild switch --upgrade" already done and rebooted the laptop also "nix-channel --update" // Sorry still new to nixos
<clever>
njha: make sure you dont lock yourself out though!
<clever>
xwza91: that sounds like a bug in nixpkgs then
<njha>
yup, I have physical access and access through wireguard
<njha>
thanks!
xelxebar_ has quit [Quit: ZNC 1.7.2+deb3 - https://znc.in]
<{^_^}>
#41212 (by Chiiruno, 2 years ago, closed): [Question] How to enable a service from a different channel
<clever>
xwza91: yep
<xwza91>
thx will try it out
gustavderdrache has joined #nixos
<xwza91>
I've "{ config, lib, pkgs, ... }:" on top, if I add channel nixpkgs-unstable with "nix-channel --add" how to import it to config to reference packages with "pkgsUnstalbe.<package>" or is this the wrong way I think?
<clever>
xwza91: you would just do <nixos-unstable> if `nix-channel --list` shows the name as nixos-unstable
<clever>
xwza91: then you should be able to just `import <nixos-unstable> {}`
<xwza91>
so I can put that on top of my configuration.nix and use nixos-unstable.<PACKAGE> in the complete file? How to find out which package is holding my old file which contains the deprecated code?
davidak_ has joined #nixos
<clever>
xwza91: its not a package, its a nixos module
<clever>
xwza91: so you want <nixos-unstable/nixos/modules/tasks/encrypted-devices.nix> in your imports list
<xwza91>
I've my complete configuration in one file and no "import " anywhere there, I'm sorry but can you give me an example how to do that?
<xwza91>
just "import <nixos-unstable/nixos/modules/tasks/encrypted-devices.nix>;" ?
<aveltras>
the install phase completes successfully
<xwza91>
clever it does not work - maybe I did somethng wrong thanks for your help so far: https://pastebin.com/06PQJ1yt (tried it also with import inside the first brackets after with lib; {)
<clever>
xwza91: imports goes after line 10, not before
jgeerds_ has quit [Ping timeout: 256 seconds]
<xwza91>
The option `fileSystems./.encrypted.enable' in `/nix/var/nix/profiles/per-user/root/channels/nixos/nixos/modules/tasks/encrypted-devices.nix' is already declared in `/root/.nix-defexpr/channels/nixos-unstable/nixos/modules/tasks/encrypted-devices.nix'.clever
fusion809 has joined #nixos
<clever>
xwza91: you didnt use disabledModules to disable the old module
<xwza91>
thx think I understand it now and will test again :)
justanotheruser has joined #nixos
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos
mbrgm_ has joined #nixos
mbrgm has quit [Ping timeout: 260 seconds]
mbrgm_ is now known as mbrgm
<xwza91>
clever yeah it's working and rebuilding now thx :) One last question, how can i install a specific package (not using a module) from unstable? e.g. firefox
<xwza91>
thx, my current "systemPackages" looks like "environment.systemPackages = with pkgs; [" can I combine them to have it only once is is it no problem to have it duplicate in code?