Robert Bragg

Since it's tangentially related, I just wanted to link this to https://github.com/rust-windowing/winit/issues/2308 - which advocates for clearer semantics for APIs like `inner_size` across platforms. In particular that would introduce the...

Just to mention that I recently (https://github.com/mesonbuild/meson/issues/3023) also found it a little awkward that `configure_file()`'s output filename couldn't include a subdirectory while I wanted to generate a `.java` file based...

I guess if you aim to have compatible support for layering on jni-sys vs direct bindgen for jni.h the main thing that would be exposed is the `jboolean` change and...

> > be exposed is the jboolean change and you could also alias jboolean as bool for bindgen for consistency? > > No, I can not see any impact of...

For reference here; I cherry picked the change to make `exception_occurred` infallible and merged that via https://github.com/jni-rs/jni-rs/pull/517 while I'm not currently 100% sure about some of the changes here.

Wow! I just clicked that link, and I don't know if I'll be able to un-see that screenshot! :exploding_head:

In general it's not expected to be a Rust/JNI safety issue to mix up object reference types (e.g. calling methods against the wrong class), since that usually leads to an...

@argv-minus-one I wonder if you might get a chance to comment here since it looks like this comes from https://github.com/jni-rs/jni-rs/pull/398

Ah, okey, phew, I wasn't entirely sure off the top of my head if it would compile to pass a `&mut JObject` here.

Track this issue/discussion around thread attachment, where it's easy to see how the use of thread-local storage across different versions of the `jni` crate could be hazardous: https://github.com/jni-rs/jni-rs/issues/441