Christian Poveda Ruiz
Christian Poveda Ruiz
I think a good compromise would be a subset of what you mention: ```rust pub type foo_ctype = ::std::os::raw::c_uint; #[repr(u32)] #[derive(Debug, Copy, Clone, Hash, PartialEq, Eq)] pub enum foo {...
> I forgot about the From/TryFrom conflict, thanks for pointing it out. Everything you said sounds reasonable to me, would you accept a PR to add that enum option? I...
> Even if a procedural macro is able to do it, `bindgen` is already a code generator, so it seems better to just perform the generation. That simplifies the build...
I think the main "blocker" here is the CLI API for these two options. The library API is more or less clear: ``` pub enum EnumVariation { Rust { non_exhaustive:...
While I understand this is annoying, there is not much we can do from the bindgen side as clang provides the definitions for these built-in types itself. So the fact...
So after a little bit more digging it seems the issue we have here is that on C `void foo(__va_list_tag arg[1]);` is just the same as `void foo(__va_list_tag *arg);`. A...
yeah but the name of the type is the same, isn't that what you wanted?
This happens because any non-trivial literal is treated as an expression so `(literal)` will trigger a clang evaluation and bindgen ignores constants that would be truncated to fit in a...
I think none of us has a windows machine to test this so it will be difficult to track down, it could very well be that clang doesn't support this...
If that's the case, bindgen passes all the arguments after `--` to clang. That might help.