gtk-rs-core macros and clippy::as_conversions

[swsnr]

Some code generated by glib-macros trips on certain clippy::restriction lints, specifically clippy::impl_trait_in_params and clippy::as_conversions (and related more specific lints about use of as). The former is probably too exotic, but personally, I do like to use the latter, because I don't like as mixing up all sorts of different safe and less safe conversions.

Currently, I'm working around this by isolating glib enums in a mod block with allow attributes, but I'd love if gtk-core-rs supported clippy::as_conversions out of the box.

But I understand that denying as is probably quite controversial, so I figured I'd put this up for discussion first.

If you agree that this would be a welcome change, I'd make a pull request to make gtk-rs-core (at least glib and gio) work without as and deny clippy::as_conversions . If you disagree, then never mind 🙂

[sdroege]

I think this would make sense but it's quite a lot of work. It's not just a matter of switching to from() / try_from() everywhere but in various places the C integer casting behaviour of as is actually what is needed (which should be no surprise as this is interfacing with a C library...) so this has to be changed very carefully. Doing that (correctly) would make the code clearer in any case.

[swsnr]

I don't think I'd want to remove all as conversions from gtk-rs-core, at least not right away. That's a lot of work, for little gain: I agree that there's little to gain in stripping as conversions from the immediate FFI boundary, and it's not something I'd ever see as a consumer of these crates.

What I'd be after first and foremost is to get rid of as conversions in macros, because these leak into downstream code.

[sdroege]

That makes sense but alternatively the macros could also emit #[allow(clippy::XXXX)] on the affected code