<multun>
It could be anything, it's probably a wine bug but that's hard to say
<mkg20001>
quinn: tried it with steam-run, I could try it with "ulimit -s 10000" to give it a bigger stack
digit_ has joined #nixos
<quinn>
that's an interesting idea, no way to know whether it'll work until you do it i guess
digit_ has quit [Client Quit]
digit_ has joined #nixos
<mkg20001>
quinn: There's a bunch more errors actually - https://paste.mkg20001.io/alofiguzab.rb - seems like it tries to open a socket, but that doesn't work, then it does a stackoverflow (regardless of raised limit)
<JJJollyjim>
and i guess at evaluation-time there's no way to know what's gonna end up in my machine closure, cos that depends on what the build outputs end up referencing
<morgrimm>
Hey all - does anyone have the gpg agent running ssh auth for them? I have it and ssh support set up, but it refuses any operation to do with my public key
<morgrimm>
Signing in git and stuff with my signing key works fine, but ssh doesn't
<{^_^}>
[nixpkgs] @bdesham opened pull request #89584 → mkvtoolnix: fix Darwin build → https://git.io/Jf1M0
<hyper_ch>
question: due to systemd there is no classic "syslog" anymore. So when giving info to a module that it logs to "syslog", should I still use "syslog" (as generic term) or should I use "systemd" or "journalctl"?
<ldlework>
Is the patchPhase I'm using sed and getting, sed: couldn't open temporary file /nix/store/2g4a5japiqmlj2xfdvqf6jqhg1ba5rnd-source/src/sedPeRDzF: Permission denied
<ldlework>
what should I do instead
<ldlework>
oh nm sorry i'm just tired
<energizer>
hyper_ch: really it's just logging to stderr, right?
<ldlework>
actually no I'm still confused -- during the patch phase, why can't I modify the source files?
<hyper_ch>
energizer: no idea :)
<hyper_ch>
energizer: I'm not all familiar with that stuff....
<hyper_ch>
If I enable the debug option, it everything shows up in journalctl
<energizer>
hyper_ch: normally systemd services print to standard streams and systemd itself puts it into the journal
<hyper_ch>
well, so far everything the old documentations all say: "if debug is enabled, everything gets logged to syslog"
<energizer>
maybe look in the source of a module and see what the debug option is doing
<hyper_ch>
energizer: I'm adding it becuase currently it's not an option
<hyper_ch>
so I need to know what to put into the helper text
<energizer>
hyper_ch: look at other ones
<hyper_ch>
since there is no syslog anymore
proofofkeags has joined #nixos
<{^_^}>
[mobile-nixos] @samueldr opened pull request #162 → Allow passing device definitions by path → https://git.io/Jf1yj
proofofkeags has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixos-homepage] @sharkbrainguy opened pull request #459 → Fix a broken link → https://git.io/Jf1SI
<clever>
jakob_rs: if the target from one of these links goes away, then it can be GC'd
<jakob_rs>
So presumably, while installing Nix on the other machine, I'd just do a normal multi-user install, and then replace /nix with the shared @nix
<clever>
yeah
<jakob_rs>
and rewrite ~/.nix-profile and ~/.nix-defexpr/channels
<jakob_rs>
In a multi-user install, can root install into a "default" profile?
<clever>
yeah
maxdevjs has quit [Ping timeout: 265 seconds]
<clever>
you could just use the same profile for both OS's
<jakob_rs>
The problem is if I have to manually keep the Nix package installed, I wouldn't want to have Nix installed in the *user profile* of the NixOS installation
<jakob_rs>
I'd imagine Nix is installed in the default profile though
<clever>
for nixos, it gets nix from /run/current-system/sw/bin/
<clever>
so you would want the default profile to be different
<jakob_rs>
yeah, so if I install Nix locally I was worried that I'd get two nix installs in nixos
<jakob_rs>
but if it's installed in the default profile that's not a problem
<clever>
you can also have seperate /nix's but use the other /nix like a read-only cache
<clever>
add local?root=/other/root/ to substituters in nix.conf
<clever>
then it will read /other/root/nix/store and copy things over to its private /nix/store
<jakob_rs>
Wouldn't that cause duplication?
<clever>
yeah
<jakob_rs>
Also I have no root/nix folder, it's located in a @nix subvolume
<jakob_rs>
but I could mount that manually
<jakob_rs>
Can Nix reflink with `local?root=/other/root1`?
<clever>
dont think so, but that might be an idea to request on nix
<clever>
there is also nix-store --optimize
<clever>
which populates /nix/store/.links/ with hardlinks of duplicate files
pjt_014 has joined #nixos
<jakob_rs>
With `local?root=`, can I specify the folder for the store specifically? Like `local?nix=/other/nix/store`
<clever>
it behaves like a chroot, so that would look in /other/nix/store/nix/store
<jakob_rs>
Can I override the nix store path inside of the `local?` uri then
<jakob_rs>
This isn't a huge problem, it just avoids an extra mount point
<jakob_rs>
So something like `local?root=/other/root&nix-store-path=@nix/store`
<clever>
ldlework: does it fail the same way if you use a local path in imports?
<jakob_rs>
I don't think you need to `import` explicitly
<ldlework>
that's what I'm doing
<jakob_rs>
{ imports = [ ./nix/sabaki ]; } without `import`
<ldlework>
OK I changed that. Still infinite recursion.
<clever>
ldlework: where does this file then get into imports?
<ldlework>
clever: in my configuration.nix in my imports
<clever>
ldlework: can you pastebin that code?
<ldlework>
just as an absolute path to this file
<clever>
ah
user_0x58 has quit [Quit: Leaving]
<ldlework>
/home/ldlework/src/baduk.nix
<ldlework>
which is a directory
<ldlework>
containing that default.nix and nix/ paths, etc
<clever>
reading the trace now
<clever>
anonymous function at /nix/store/lcnb1m4f4h3zm8v3b287l5sysqwh33w9-nixos-20.09pre226148.0f5ce2fac0c/nixos/pkgs/build-support/build-fhs-userenv/env.nix:3:1 called with unexpected argument 'version', at /nix/store/lcnb1m4f4h3zm8v3b287l5sysqwh33w9-nixos-20.09pre226148.0f5ce2fac0c/nixos/pkgs/build-support/build-fhs-userenv/default.nix:8:9
<clever>
thats not infinite recursion
<jakob_rs>
Are you sure that the github repo is synced?
<jakob_rs>
I don't see any references to the package mentioned in the trace in default.nix
proofofkeags has quit [Ping timeout: 246 seconds]
slack1256 has quit [Remote host closed the connection]
<ldlework>
hmm i must've pasted the wrong traceback or something strange
<ldlework>
while evaluating the module argument `pkgs' in "/home/ldlework/src/baduk.nix/nix/sabaki":
<ldlework>
Oh I think I just figured it out.
<ldlework>
nope :(
* clever
reads latest trace
<ldlework>
Just fixed a small error where the enableOption was being assigned to `options.baduk.sabaki` instead of `options.baduk.sabaki.enable` but that didn't fix the recursion
<clever>
ldlework: try using `{ config, pkgs, lib, ... }:`
<clever>
and then `with lib;`
<ldlework>
clever: why?
<ldlework>
I used to do that
<ldlework>
I do that in my personal nixos modules, but I removed that for these public ones
<ldlework>
because I couldn't remember why I was accepting lib
<clever>
because nix needs to know what your config= on line 11 returns, before it knows what the value of pkgs is
<ldlework>
and I looked in the docs, and it doesn't even mention that modules are sent `lib`
<clever>
config= depends on pkgs.lib.mkIf
<clever>
pkgs depends on config.nixpkgs.overlays
<clever>
infinite recursion
xO1 has joined #nixos
<ldlework>
damn i wish i could genuinely grok that
<ldlework>
that worked though
nootnoot has joined #nixos
<nootnoot>
noot
<ldlework>
lol
<nootnoot>
:)
<meh`>
noot!
<nootnoot>
meh`
<nootnoot>
you're a smart man
<nootnoot>
I feel like you and I will get along just fine
<meh`>
:D
<nootnoot>
:D
<nootnoot>
How's everyone doing
<jakob_rs>
I've now given up looking for where `root=` is defined
o1lo01ol1o has joined #nixos
<clever>
jakob_rs: its deep in the source for `class LocalStore` i think
<jakob_rs>
It seems that this "parameter handling" is split between the various store implementations
<meh`>
nootnoot, I'm alright, trying to get GitLab to work properly, failing, how about you?
wavirc22 has quit [Quit: ZNC 1.7.2+deb3 - https://znc.in]
<jakob_rs>
No wait actually I did find 'root=', just no others.
<pjt_014>
hi! I'm trying to use this thing ( https://github.com/illegalprime/nixos-on-arm ), but the nixpkgs submodule is months out of date and is at version 18/09 or something.
<pjt_014>
(How) can I update it while keeping the arm tweaks intact?
<pjt_014>
git submodule --help doesn't really make it clear. If I want to be sure, can I make/reapply a patch or something?
<jakob_rs>
It seems it depends on a fork of nixpkgs
<pjt_014>
Yeah, the fork has some fixes for arm boards that still need to be upstreamed
wavirc22 has joined #nixos
<jakob_rs>
you could try to rebase the fork onto master
<jakob_rs>
it looks like it's about 30 commits off upstream from the git log
o1lo01ol1o has quit [Ping timeout: 246 seconds]
juhe has joined #nixos
<ldlework>
Can someone remind me, if I have some modules, how to generate documentation for them?
<{^_^}>
[nixpkgs] @FRidh pushed 1000 commits to python-unstable: https://git.io/Jf1Hn
<redcedar[m]>
I have an install question: I seem to successfully get through the install process (I was able to boot into root without a display manager), but when I reboot with a new user and xserver enabled it hangs with "Failed to start X11 server" after starting and stopping it several times. The problem is that at this point I no longer have access to a console, and cannot debug the failed x11 server (even though I was able to boot
<redcedar[m]>
into a display manager during the install). Any ideas on what I should do?
<{^_^}>
[nixpkgs] @danieldk opened pull request #89607 → pythonPackages.pipBuildHook: do not build in an isolated environment → https://git.io/Jf1QY
fendor has joined #nixos
amerigo has joined #nixos
zupo has quit [Ping timeout: 272 seconds]
drewc has quit [Ping timeout: 272 seconds]
magnetophon has quit [Remote host closed the connection]
zupo has joined #nixos
<redcedar[m]>
juhe: yes that worked! I ran journalctl and the more precise error just before failing after 3 attempts is "Start request repeated too quickly"
<juhe>
redcedar[m]: great!
eoli3n_ has quit [Ping timeout: 256 seconds]
davidv7 has joined #nixos
asymptotically has joined #nixos
xO1 has joined #nixos
est31 has quit [Remote host closed the connection]
est31 has joined #nixos
proofofkeags has joined #nixos
drewc has joined #nixos
proofofkeags has quit [Ping timeout: 260 seconds]
<redcedar[m]>
juhe: thanks, I was able to get it working =)
<juhe>
redcedar[m]: :-) glad the provided info helped
<jluttine[m]>
Have you used NixOS on Intel NUCs? (For instance, https://www.intel.com/content/www/us/en/products/boards-kits/nuc/kits/nuc7pjyh.html) Do you know if it's likely to work similarly as any other desktop or is it somehow so different that one should expect problems and difficulties? For instance, is it compatible with `x86_64`..
<jluttine[m]>
* Have you used NixOS on Intel NUCs? (For instance, https://www.intel.com/content/www/us/en/products/boards-kits/nuc/kits/nuc7pjyh.html) Do you know if it's likely to work similarly as any other desktop computer or is it somehow so different that one should expect problems and difficulties? For instance, is it compatible with `x86_64`..
quinn has quit [Ping timeout: 265 seconds]
<etu>
jluttine[m]: x86_64 is also known as amd64, it's the same architecture that almost every non-phone/non-tablet that is used by people out there is using.
Jackneill has quit [Ping timeout: 246 seconds]
<jluttine[m]>
etu: yeah. i was just wondering because simple single-card computers like raspberry pi use arm, so i thought if NUCs use ARM too.. but i guess not
<jluttine[m]>
i don't know almost anything about these..
Jackneill has joined #nixos
<jluttine[m]>
but yeah, seems to be x86
<etu>
jluttine[m]: Well, intel want to compete with the smaller computers as well ;)
<jluttine[m]>
so should be no problems with nixos in that sense
<jluttine[m]>
etu: intel doesn't do ARM at all?
<etu>
Not what I know of :)
hmpffff has quit [Quit: nchrrrr…]
<{^_^}>
[nixpkgs] @flokli merged pull request #86994 → wire-desktop: linux 3.17.2924 -> 3.18.2925, mac 3.17.3666 -> 3.18.3728 → https://git.io/JfZmu
<{^_^}>
[nixpkgs] @flokli pushed 3 commits to release-19.09: https://git.io/Jf1dk
<sherub>
Hi folks I am using a tmux plugin which is not in nixpkgs and wanted to contribute (i.e. add that plugin to nixpkgs) but I am not really sure on how do I test out the changes I have made. Is there any comprehensive newbie guide on testing changes to nixpkgs?
<{^_^}>
[nixpkgs] @danieldk opened pull request #89624 → doc: `pip install` flags are specified with `pipInstallFlags` → https://git.io/Jf1dX
<sherub>
It is a fairly small change at the moment, but still would like to understand the flow, and how to get it merged quickly. https://github.com/NixOS/nixpkgs/pull/89623 is the pull-request I would like help in testing.
amerigo has quit [Quit: Connection closed for inactivity]
gulplante has joined #nixos
<{^_^}>
[nixpkgs] @teto closed pull request #78565 → Lua for sile → https://git.io/JvqNq
tobiasBora2 has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<tobiasBora2>
Hello, I'd like to know, my nixos is configured with static IP (because it's supposed to be a server), however I changed the box recently and the gateway is wrong now, so it can't access internet. And the problem is that when I run "nixos-rebuild", it tries to fetch some packages from the internet... Any idea if I can ask him to first fix the internet before fetching the packages?
<hyper_ch>
it builds with python 2.7... shouldn't you use python 3?
<tobiasBora2>
hyper_ch: for now I'm using it for me and I'd prefer the latest version, after if people in nixpkgs are interested by it sure I'll use the releases ;-)
<tobiasBora2>
the error being ModuleNotFoundError: No module named 'dyn_gandi'
<tobiasBora2>
(and indeed in the source code all the dyn_gandi.py code has been removed, but I'm not good enough in nix/python packaging to understand the issue
<hyper_ch>
tobiasBora2: no idea.... I don't like pythong much
<magnetophon>
Where should a makefile install /etc/ so that it just works on NixOS? $(DESTDIR)/etc/ ?
dingenskirchen1 is now known as dingenskirchen
zimbatm_ has joined #nixos
<{^_^}>
[nixpkgs] @oxalica opened pull request #89649 → rust-analyzer: 2020-04-20 -> 2020-06-01 and fix build of vscode extension → https://git.io/Jf1xN
<jlv[m]>
I'm trying to disable `enableACME` for every `nginx.virtualHosts.<name>`, but I keep running into infinite recursion errors. My code looks like `services.nginx.virtualHosts = builtins.listToAttrs (map (n: { name = n; value = { enableACME = false; }; }) (builtins.attrNames config.services.nginx.virtualHosts));`. Theoretically, this should be possible, because `attrNames config.services.nginx.virtualHosts` does not change, but I
<jlv[m]>
don't know how to make the compiler understand that.
<morgrimm>
I can sign things properly, but it seems like the agent is just refusing any SSH operations, despite the socket being set to the agent socket
<elvishjerricco>
morgrimm: Using GPG as your SSH agent has always been kinda messed up. It has to do with the gpg-agent not knowing which TTY the user is actually using, and SSH providing no way for the agent to ask the SSH process. There's hacks that can improve the situation but I never could get them to work properly in all the cases I needed.
<{^_^}>
[nixpkgs] @mmilata opened pull request #89668 → lnd: 0.10.0 -> 0.10.1, enable same features as upstream → https://git.io/JfMvG
<Thra11>
Do we currently have any mechanism for linking issues to maintainers (e.g. if someone reports a problem with somepackage, is there a way for the somepackage maintainer to get a notification automatically)?
<joko>
Thra11: There's at least some mentioning in the Issue Template about notifying the package manintainers
<morgrimm>
joko: Is this in a home manager config? I'm using that exact config for it, but I had to add some env vars to my shell to get the correct socket - did you do any shell profile changes to support it?
<simpson>
...Fair enough, I guess I shouldn't ask the bot to evaluate packages.
gxt has quit [Ping timeout: 240 seconds]
<kaliumxyz>
and then use sed to change the makefile in the build script to use an absolute path to the pkgconfig binary instead of it trying to get it from path?
urkk_ has joined #nixos
urkk has quit [Ping timeout: 265 seconds]
<simpson>
Sure, although I was thinking of just arranging the environment to satisfy the Makefile.
<kaliumxyz>
the makefile wants the library in the path.
<kaliumxyz>
I guess because it switches shell its not there?
urkud has joined #nixos
gxt has joined #nixos
<urkud>
Hi, is there an easy way to install old TeXLive from nixpkgs?
artemist has joined #nixos
wnklmnn has joined #nixos
dhess has quit [Remote host closed the connection]
<urkud>
Publisher has TL2013, they want me to upload the source files of my paper, and I can't generate a biblatex .bbl file compatible with their TL.
<simpson>
urkud: Not really. Are you running into the problem where the old tarballs aren't available anymore? AIUI upstream simply doesn't have the tarballs any longer.
<simpson>
Oh no, that's a terrible problem. I'm not sure what can be done.
<evanjs>
I did _not_ have such a strong feeling of involvement with Gentoo. Especially after the whole mailing list whitelist thing years ago. Not that I was super involved, it just felt... idk
meh` has quit [Ping timeout: 246 seconds]
<gchristensen>
evanjs: isn't it amazing?
<evanjs>
It's also just so much easier to contribute to nixos
<Yaniel>
oh, since this channel is active, anyone familiar with using appimages on nixos?
<Yaniel>
I have one that fails to launch due to not finding icu4c
<evanjs>
gchristensen: indeed it is. I also love being able to contribute without always fully committing to something (I know maintainers are good but bah)
<{^_^}>
[nixpkgs] @bhipple pushed to master « mu: add changelog link in meta »: https://git.io/JfMfd
<evanjs>
Yaniel: I haven't used them much, but to clarify, is this an expression that handles an appimage? Or is this a naked appimage that isn't handled by any nix expressions?
amanjeev has quit [Quit: Logging off! Bbye!]
<Yaniel>
no a non-nixpkgs appimage
<evanjs>
😰
<{^_^}>
[nixpkgs] @teto merged pull request #89632 → lyaml: init at 6.2.5-1, update almost all luarocks generated packages → https://git.io/Jf1FD
<{^_^}>
[nixpkgs] @teto pushed commit from @lblasc to master « lyaml: init at 6.2.5-1, update all generated packages »: https://git.io/JfMfb
<Yaniel>
that's the reaction I was afraid it would get :D
<evanjs>
Afaik that probably won't work without wrapping
<evanjs>
It's pretty straightforward and, depending on the package, might not be too difficult. I had to do it for UAExpert... which I should probably open a PR for already omg
<kaliumxyz>
Progress has been made simpson c: So there is no pkg config file in libusb1, yet pkg-config expects this (the hook mangles the make script if I include it seemingly leaving libusb fields empty?)
<ar>
is it just me, or buildLinux (pkgs/os-specific/linux/kernel/generic.nix) doesn't use multiple jobs for building kernels?
<tobiasBora2>
I don't know why but on my rasp, the time service can't start. I have an error: "systemd-timesyncd.service: Failed to set up special execution directory in /var/lib: Not a directory"
<tobiasBora2>
Ok, this link suggest to remove a symlink
<{^_^}>
systemd/systemd#12131 (by kouros17, 1 year ago, closed): RFE: migrate StateDirectory= and friends back from DynamicUser=1 subdir to regular one if needed
<tobiasBora2>
I'll try
<kaliumxyz>
ah I figured it out, its because of the author of the makefile includes some git data into the program for versioning.
<ar>
nvm, that was just me. one of the phases was running without -j16, and it got more evident because of running under binfmt emulation…
<tobiasBora2>
great, problem solved, sorry
<tobiasBora2>
It's not clear to me, during a "nixos-rebuild switch", if some services fail, will it switch or not?
KindTwo has joined #nixos
o1lo01ol1o has joined #nixos
KindOne has quit [Ping timeout: 272 seconds]
<Thra11>
tobiasBora2: I believe it does switch.
<tobiasBora2>
Thra11: ok thanks. Then is there a way not to switch if a service fail?
<tobiasBora2>
(ideally re-spanning the old services)
KindTwo is now known as KindOne
<Thra11>
tobiasBora2: Not that I'm aware of. You can always do `nixos-rebuild --rollback switch` if you see failures which will stop the new config working.
<tobiasBora2>
Thra11: ok thanks
o1lo01ol1o has quit [Ping timeout: 272 seconds]
<{^_^}>
[nixpkgs] @primeos pushed to master « chromiumDev: Fix the configuration phase »: https://git.io/JfMUl
<betaboon>
I'm running into "...armv7l-unknown-linux-gnueabihf/bin/cmake: cannot execute binary file: Exec format error" when trying to cross-compile certain packages for armv7, this block ceveral packages from being cross-compiled. anyone knows anything ? I'm somewhat lost here
<Thra11>
betaboon: If I had to take a wild stab in the dark, maybe these certain packages have cmake in `buildInputs`, not `nativeBuildInputs`? Do you have an example of a broken package?
mallox has quit [Quit: WeeChat 2.8]
magnetophon has quit [Remote host closed the connection]
<betaboon>
Thra11: when the package had cmake in buildInputs it couldnt find cmake at all, when i added cmake to nativeBuildInputs the mentioned error occured. i just now tried adding `buildPackages.cmake` to nativeBuildInputs -> that made the package work oO
<energizer>
perhaps you can patch the functools-lru-cache backport to delete that file
fendor has joined #nixos
<energizer>
fwiw cpython 2 is past end-of-life and matplotlib doesn't support it anymore
<rooke>
Yeah, I've been asked to build a code base which uses it
<rooke>
I'm not thrilled about it lol
urkk has joined #nixos
Thra11 has quit [Ping timeout: 258 seconds]
<energizer>
i can explain what's going wrong, and probably guess how to fix it, i'm just not familiar enough with the nix conventions to know what the idiomatic way to fix it is
urkk_ has quit [Ping timeout: 246 seconds]
<hexa->
,locate far.h
<{^_^}>
Found in packages: far2l
<rooke>
I'd appreciate it, I don't really know whats going wrong beyond the obvious some file is trying to be placed where another file already lives
<rooke>
Also on the matplotlib support, it looks like nix checks if you're using python two and loads an older matplotlib in that instance
LevelChart8 has joined #nixos
<energizer>
you have installed two backports (modules from python3 made avaialble as third-party packages for python2), one providing ssl functionality and one providing the py3 functools.lru_cache. both of those modules are using the `backports` namespace and create a backports/__init__.py file. normally python wouldn't care but nix doesn't like collisions
dingenskirchen has quit [Remote host closed the connection]
<rooke>
I just kept searching for matplotlib based issues
Vikingman has joined #nixos
pjt_014 has joined #nixos
<iqubic>
I have a few pulseaudio loopbacks and null-sinks created. Is there a way to set which streams have their sound levels manipulated when I set "sound.mediaKeys.enable = true" in my configuration.nix?
<lovesegfault>
flokli: Maybe
<lovesegfault>
Not sure I want to fight the good fight against binutils any more
<lovesegfault>
Bumping it breaks _everything_
<ldlework>
Is there a way to have a listOf enum, such that each enum value can only be passed once?
xcmw has joined #nixos
<energizer>
ldlework: you could encode it as Dict[str, bool]
<energizer>
known keys, bool values
<ldlework>
energizer: the user could provide any string then right?
<energizer>
ldlework: i mean just like a regular module that accepts certain options
<energizer>
bool-valeud
<energizer>
-valued
<ldlework>
right, but the appeal of an enum is that nix enforces the values from a set of approved ones
<energizer>
instead of [ "foo" "bar"] it's foo = true; bar = true;
<ldlework>
so the user could do doesntexist = true;
<energizer>
it already does taht. if you pass non-existing keys to a module it'll complain
<ldlework>
ohhhhhh
<infinisil>
This doesn't throw an error for invalid names: `type = attrsOf bool`
<infinisil>
I think energizer is talking about `type = submodule { options.foo = mkOption { ... }; options.bar = mkOption { ... }; }` though
<energizer>
yes
gxt has quit [Ping timeout: 240 seconds]
gxt has joined #nixos
<ldlework>
right thank you
tertl3 has joined #nixos
<{^_^}>
[nixpkgs] @bouk opened pull request #89693 → Add gopls to vim-go binaries → https://git.io/JfMLy
<{^_^}>
[nixos-hardware] @mexisme opened pull request #170 → Initial port of linux-surface/linux-surface patches to NixOS → https://git.io/JfMq8
magnus has joined #nixos
karantan has joined #nixos
<energizer>
iqubic: are you interested in packaging it yourself?
thmzlt has quit [Remote host closed the connection]
<iqubic>
I honestly have no idea how to do that.
<energizer>
iqubic: are you interested?
<iqubic>
I am indeed.
<energizer>
in learning to do it
<energizer>
oh it already has a default.nix
magnus is now known as skorpy
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<iqubic>
But I'm going to be busy for the next hour or so. I'll take a look at doing this later.
zzywysm has joined #nixos
<Vikingman>
someone here managed to get retroarch working ?
<ldlework>
I think so
dermetfan has joined #nixos
<ldlework>
Oh I have a "retroshare" module
<ldlework>
lol sorry
<zzywysm>
according to commit 6599499cd7793b085edd0072b7bbde4039626753 the nixos 20.03 channel should be on linux 5.4.44. so how come my nixos installation is running linux 5.4.35?
<energizer>
unless something's pinning Pillow==7.1.2 i dont see how that could even happen
<adisbladis>
energizer: Huh? What do you mean?
<energizer>
adisbladis: some application says setup(python_requires='2', install_requires=['pillow']), pip should give a version of pillow that works with py2
<energizer>
and afaik it does do that
<adisbladis>
No, it doesn't.
<adisbladis>
Ah, right. _pip_ does that.
<adisbladis>
Nixpkgs python packaging does not, all this metadata from the python package needs manual intervention & manual copying
m0rphism has joined #nixos
<energizer>
is it possible to automate the transition to poetry2nix in nixpkgs?
<energizer>
adisbladis: currently nixpkgs doesn't use poetry2nix for most python packages, but it could
<adisbladis>
energizer: Sadly it couldn't :/
<adisbladis>
What's in pythonPackages is not really solvable
<energizer>
adisbladis: oh, how come?
<adisbladis>
You couldn't put it all in a single big environment and actually solve the environment correctly
<energizer>
i wasnt thinking they'd all go in the same pyproject.toml
<adisbladis>
Though, why would you even want to?
<energizer>
but each package could get its own
<adisbladis>
Yes, that makes sense
<energizer>
the advantage is that hydra would be building all the packages instead of me doing it :)
<energizer>
also probably easier to maintain
<DigitalKiwi>
colemickens: lol i just got that too trying nixpkgs master because i couldn't build from nixos-unstable because whatever tor-browser (i think tor-browser-bundle-bin) doesn't build because of some python zope...
<b42>
i'm trying to buildPythonApplication but the pipInstallPhase fails with ERROR: Could not find a version that satisfies the requirement dbus-python>=1.2 (from dhcpcanon==0.8.5) (from versions: none)
<DigitalKiwi>
which seems to be the only tor-browser these days?
<b42>
the dbus-python package has no dist-info directory which seems weird but there are plenty derivations in nixpkgs that use dbus-python and build fine, what am i missing?
<DigitalKiwi>
that is not the one that's making pillow error though :(
<colemickens>
DigitalKiwi: I finally spotted calibre in the --show-trace and that aligns with something about python2 and deprecation and eh, I'll probably just remove it from my system and move on
<energizer>
b42: if i had to guess sounds like dhcpcannon needs .overridePythonAttrs(old: {propagatedBuildInputs = [self.dbus-python];}) and you'll have to define self.dbus-python somewhere too
<DigitalKiwi>
ah, scribus
<DigitalKiwi>
shit i think i'm a maintainer
<adisbladis>
Don't worry, there is an open PR fixing Pillow
<adisbladis>
I'm testing it right now
<DigitalKiwi>
scribusUnstable specifically though idk about scribus less unstable
<DigitalKiwi>
same
<b42>
energizer: i have dbus-python in propagatedBuildInputs and its defined in pythonPackages: https://paste.rs/E1h
<DigitalKiwi>
so if you fix pillow you fix at least 3! good deal
<energizer>
b42: what's the version in pythonPackages?
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<DigitalKiwi>
adisbladis: lol and nixops
<adisbladis>
Yeah, I'm a bit grumpy about this being merged...