17 Oct 2024 |
K900 | So wlroots can say private.requires: /nix/store/aaaaaaaa-libffi/lib/pkgconfig/libffi.pc | 10:46:51 |
emily | I wonder if pkgconf will accept that. | 10:47:02 |
K900 | But that's insane | 10:47:06 |
emily | it would be easy to do the patching in fixup | 10:47:08 |
K900 | As in an insane amount of work | 10:47:11 |
emily | not insane if we automate the patching | 10:47:18 |
emily | have something in fixup that parses the requires stuff and rewrites to use paths | 10:47:32 |
emily | it's pretty comparable to shebang patching right? | 10:47:56 |
K900 | I guess | 10:48:11 |
Alyssa Ross | i am modifying the perl script that generates the xorg packages, which analyses their .pc files to find build inputs, to propagate them | 10:49:13 |
Alyssa Ross | I am in hell | 10:49:15 |
emily | lol | 10:49:27 |
K900 | @Artturin what happened to your X11 PR again | 10:49:36 |
K900 | We had a PR to just build them NORMALLY | 10:49:44 |
emily | are you sure you don't just want K900 to restore the hack but better for now | 10:49:45 |
K900 | And then it got lost somewhere | 10:49:48 |
emily | if we can potentially do .pc rewriting to use absolute paths next cycle | 10:49:56 |
emily | which would be nicer than propagation | 10:49:59 |
K900 | We can't, we need pkg-config to support that | 10:50:15 |
K900 | And not next cycle probably but one after | 10:50:20 |
K900 | At the earliest | 10:50:30 |
emily | by next cycle I mean release cycle | 10:50:45 |
emily | I am implicitly volunteering to look at patching pkgconf to support absolute paths if necessary and writing some script to do the .pc patching for 25.05 | 10:51:05 |
emily | which I may regret, but hey | 10:51:14 |
emily | it means nobody has to touch any Perl at least | 10:52:01 |
Alyssa Ross | oh the perl part is fine | 10:52:38 |
Alyssa Ross | it's just find and replace buildInputs with propagatedBuildInputs | 10:52:46 |
Alyssa Ross | it's the fact that I find myself doing this in the first place | 10:53:02 |
emily | it still seems better to avoid doing the work of propagation now if our ideal end-state involves absolute .pc -to-.pc references rather than propagation | 10:53:09 |
emily | for all our sakes | 10:53:18 |