atk: clean up cross-compilation logic#124417
Conversation
|
I of course find first claim unreliable and the second change net negative. You might be able to convince @jtojnar to merge it, I guess. |
|
Keeping it like it was is fine by me, I don't have a strong opinion on either way. I'll leave this open for a bit so we can decide on how we'll do this. |
8da5f61 to
0e81850
Compare
|
Having some issues where atk doesn't want to build anymore on my pc. It's hard to debug where the issue lies. Also happens on master (by forcing a rebuild using I don't understand what's happening. No idea |
|
Ah, that's probably missing /bin/sh in the sandbox |
|
Yes indeed, thanks! Doing something like this works: CaptainSpof/dotfiles@ad425fd |
This should be solved by a fixShebang on the file. |
|
@SuperSandro2000 it was also failing on master, probably due to this: #124215 So that fail is not related to this patchset. |
|
I'm not sure if we should move forward with this or just close it. I wanted to make this package more consistent with the other PRs that were applied, but the result of applying this or not will be the same, except for the looks. If this doesn't move, I think I'll just close it in a week or so. |
Motivation for this change
Clean up based on review comments on other packages. See e.g. #123712.
Remove conditionally adding gobject-introspection as nativeBuildInput (it doesn't matter if it's available or not), remove input parameter that can just be inlined.
I think it's nicer to have the cross support be as minimal as possible to have a clean package definition.
Things done
sandboxinnix.confon non-NixOS linux)nix-shell -p nixpkgs-review --run "nixpkgs-review wip"./result/bin/)cc @7c6f434c