treewide: remove issue #56943 workarounds#215715
Conversation
e7be096 to
fc9640a
Compare
checked with diffoscope+difflog and launching the programs i have fixed that issue in my other gir PRs
and fix double wrapping
it still fails to launch though but its not a regression
$ ./result/bin/dogtail-logout
Traceback (most recent call last):
File "/nix/store/az8mw2pmhbwlm08176wd62hn0hfzwv8s-python3.10-dogtail-0.9.11/bin/.dogtail-logout-wrapped", line 8, in <module>
from dogtail.tree import *
File "/nix/store/az8mw2pmhbwlm08176wd62hn0hfzwv8s-python3.10-dogtail-0.9.11/lib/python3.10/site-packages/dogtail/tree.py", line 5, in <module>
from dogtail import predicate
File "/nix/store/az8mw2pmhbwlm08176wd62hn0hfzwv8s-python3.10-dogtail-0.9.11/lib/python3.10/site-packages/dogtail/predicate.py", line 6, in <module>
from gi.repository import GLib
ModuleNotFoundError: No module named 'gi'
pygobject3 has to be propagated for the import line in the python file
in bin
$ ./result/bin/curtail
Traceback (most recent call last):
File "/nix/store/fqjld9nchfwqd9x1pywmjfdsqrk4lxds-curtail-1.3.1/bin/.curtail-wrapped", line 37, in <module>
import gi
ModuleNotFoundError: No module named 'gi'
fc9640a to
ab3ba31
Compare
|
Actually, we might want to add some tests to https://github.com/NixOS/nixpkgs/blob/master/pkgs/build-support/setup-hooks/wrap-gapps-hook/default.nix |
already did for gdk-pixbuf in https://github.com/NixOS/nixpkgs/pull/208537/files#diff-fa727d655e0d18c0ba8bb953613ed08fb7853f04b226341ee2fe6b17957cf523R74 for gobject-introspection there's already |
|
Those do not have |
NixOS/nixpkgs@
the ones using EDIT disregard, they're testing the package which was not built with strictDeps, i'll add strictDeps |
luckily the lib output does not contain binaries therefore its safe to remove it from propagatedBuildInputs
'nix build -f . "wrapGAppsHook.tests"' pass
|
alright |
for typelibs and girs
|
due to nixpkgs/pkgs/development/libraries/librsvg/default.nix Lines 121 to 126 in 2ea8b00 |
|
Would not it make more sense to keep the test failing until it is properly fixed (e.g. using emulator to generate the loaders), rather than mask the broken runtime behaviour? |
e04770b to
d5b6b62
Compare
fixed it |
|
|
||
| # Not generated when cross compiling. | ||
| postInstall = lib.optionalString (stdenv.hostPlatform == stdenv.buildPlatform) '' | ||
| postInstall = lib.optionalString (stdenv.hostPlatform.emulatorAvailable buildPackages) '' |
There was a problem hiding this comment.
After this PR, I also need to add this diff for librsvg to build with cross again:
diff --git a/pkgs/development/libraries/librsvg/default.nix b/pkgs/development/libraries/librsvg/default.nix
index 9974826573a..87ddcd28403 100644
--- a/pkgs/development/libraries/librsvg/default.nix
+++ b/pkgs/development/libraries/librsvg/default.nix
@@ -147,11 +147,11 @@ stdenv.mkDerivation (finalAttrs: {
mv $GDK_PIXBUF/loaders.cache.tmp $GDK_PIXBUF/loaders.cache
mkdir -p "$out/share/bash-completion/completions/"
- $out/bin/rsvg-convert --completion bash > "$out/share/bash-completion/completions/rsvg-convert"
+ ${stdenv.hostPlatform.emulator buildPackages} $out/bin/rsvg-convert --completion bash > "$out/share/bash-completion/completions/rsvg-convert"
mkdir -p "$out/share/zsh/site-functions/"
- $out/bin/rsvg-convert --completion zsh > "$out/share/zsh/site-functions/_rsvg-convert"
+ ${stdenv.hostPlatform.emulator buildPackages} $out/bin/rsvg-convert --completion zsh > "$out/share/zsh/site-functions/_rsvg-convert"
mkdir -p "$out/share/fish/vendor_completions.d/"
- $out/bin/rsvg-convert --completion fish > "$out/share/fish/vendor_completions.d/rsvg-convert.fish"
+ ${stdenv.hostPlatform.emulator buildPackages} $out/bin/rsvg-convert --completion fish > "$out/share/fish/vendor_completions.d/rsvg-convert.fish"
'';
postFixup = lib.optionalString withIntrospection ''Is this the correct approach for fixing that?
If so, I'll open a PR to fix librsvg cross (I'll probably use a let rsvg-convert = "${stdenv.hostPlatform.emulator buildPackages} $out/bin/rsvg-convert" in ... to shorten it a little, though)
There was a problem hiding this comment.
Ah, thanks! I did a quick search but missed that. Sorry for the noise 😅
checked with diffoscope+difflog and launching the programs
i have fixed that issue in my other gir PRs###### Description of changes
Things done
sandbox = trueset innix.conf? (See Nix manual)nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage./result/bin/)