Mark Hammond
Mark Hammond
m-c is still on 0.5. Is it worth landing a semver range? Let's us close this and works when the inevitable update does happen...
The metadata created by different platforms should be identical - I agree we should promise it in the docs tbh. I wonder how we could test this?
as I was updating the conditional compilation docs, I thought about this. The caveat here is conditional compilation and the fact different APIs might be shipped to different platforms as...
> If user code that depends on uniffi has its own conditional compilation, that of course couldn't be expected to work in a general way. That is what I meant....
I think this is ready to roll!
This is also in #6289, so let's kill this and land it all together.
I think I got most of the review comments, but still a little unclear on how to proceed re the schema
> References in traits (and functions) feel tricky: they're definitely more ergonomic for Rust callers, but they hide a "surprise copy" for foreign callers. I think most folks would expect...
But we can have both, right? We should be able to sniff out a `Cow` and carry a flag in the metadata to the bindings?
> It was not obvious until much later that uniffi cloned everything, and I think that it's bad to allow "fake" references in the first place. To be clear, you...