Xavier Claessens

> Because we’d have to treat cargo as always stale which makes incremental rebuilds really slow I don't think that's true, it would invoke cargo which would be no-op, unless...

> But most importantly, calling one build system from another is super unreliable and cumbersome I don't think so, GStreamer-rs has been doing that with 90 lines script https://gitlab.freedesktop.org/gstreamer/gst-plugins-rs/-/blob/main/cargo_wrapper.py. Compaired...

Also, how do you deal with build.rs, my understanding was that we cannot support crates that use that, can we? Are those crate rare enough?

OOC, for mesa specifically, which crates are they going to depend on? If it's just a few selected crates, they could have native meson support just like we do with...

@sdroege I changed my mind since that old comment and I agree with you. Especially since your recent experiment, I totally support this PR, at least in the concept. I...

It's missing doc and release notes, but I like the feature. I expect some people will want to get all options instead of just user defined options, but that can...

Right, that's a known (I guess only by me) limitation of the per-subproject compiler options, because from an implementation POV they are different. We definitely should fix this, but it...

This was motivated by https://gitlab.gnome.org/GNOME/glib/-/merge_requests/3077#note_1595625

I'm wondering if we should instead extend `depends` kwarg to allow files/strings and deprecate `depend_files` from `custom_target`. IMHO it makes little sense from an API pov to split them. It...

See https://github.com/mesonbuild/meson/pull/11061