Dmitry Timofeev

> they are extremely unsafe and may turn in NULL at any moment. They seem to be about as unsafe as usual WeakReferences: you first need to turn them into...

I think it wasn't implemented earlier just because of lack of demand for this feature :-) One thing regarding the implementation: given its nature and drastic difference from GlobalRef (can...

@johnwalker , sorry for the massive delay! I'll try to review the updates and the next chapters in the coming weeks. I've recently switched jobs, and no longer work for...

Thanks a lot! Do you need a board to keep track of any smaller tasks on this tutorial (like this one: https://github.com/jni-rs/jni-rs/projects/1)? It might be more convenient than a checklist...

Sure: https://github.com/jni-rs/jni-rs/projects/2

Weird; I guess the "Triage" role is not enough; please try with "Write". I've shared the feedback to GH team.

Thanks for the report! The throughput improvement here would certainly be welcome. Part of the difference might come from using different methods: `NewObject` in C and `NewObjectA` [in Rust](https://github.com/jni-rs/jni-rs/blob/master/src/wrapper/jnienv.rs#L940). Here...

Also, some benchmarks we have might help: https://github.com/jni-rs/jni-rs/blob/master/CONTRIBUTING.md#run-benchmarks

:+1: Have you managed to confirm that the difference comes from using `NewObjectA` (not mainly from checking for exceptions, which is required by the spec and which we cannot change)?

Thanks for the issue! @vitvakatu , @DarkEld3r do you agree with marking these as unsafe? --- Regarding the use case for these methods, it appears they support manipulating a 'nativeHandle:...