Emilio Cobos Álvarez
Emilio Cobos Álvarez
Ah, nice find! Yeah, it's definitely a bug that `Some(RenameRule::None)` and `None` behave differently. Would you be willing to send a patch for this?
Yes, that sounds good!
Not really, though it added the infrastructure to potentially do it.
This is basically the same root cause as #326 (basically, that we don't simplify further once you have a "named" type, be it an alias or an struct). But yeah,...
Seems reasonable to me. Part of the issue is that `Box` is not the same size as a real pointer, so you're double-boxing which seems unfortunate. I think the hard...
Hmmm, which c compiler / library are you using? We compile all our tests to verify that they are valid C / C++ code, and we do have test-cases for...
Do the cbindgen tests pass for you there? I'm surprised since I use glibc and gcc9 locally and it works. Anyhow probably trivial to just expand the set of default...
So this is because cbindgen isn't able to follow paths, so it gets confused, and your MyStruct in ffi.rs is indeed not generic...
I _think_ you can hack around this, fwiw, with something like: ``` // ffi.rs pub use super::MyStruct as MyGenericStruct; type SomeType = MyGenericStruct; ```
Should we do this now?