Marijn Suijten

I don't really like that approach as it's quite unsafe (ie. allows users to break things at will), requires careful consideration by the caller on what `ndk-glue` version the `static`...

> but your last release made it already deprecated. openxrs also breaks if I update ndk-glue to 0.6 in my project. This is not sustainable. Indeed, and increased use of...

> @MarijnS95 In the end, is the best approach is to set `ndk = "*"` and `ndk-glue = "*"` in libraries that use android-ndk? It's such a pain to bump...

@msiglreith > I'm currently leaning towards the separate ptr crate approach as it seems to minimize the risk of breaking changes compared to just using a version range for `ndk`....

I think we should close this issue as soon as most popular crates migrated to `ndk-context`: https://crates.io/crates/ndk-glue/reverse_dependencies - https://github.com/app-dirs-rs/app_dirs2/pull/19 - https://github.com/Ralith/openxrs/pull/108 - https://github.com/RustAudio/cpal/pull/641 - https://github.com/katyo/oboe-rs/pull/42 - `webbrowser` - `winit` (needs...

> Here I used `unreachable!()` to avoid adding an unused variant in `MediaCodecOutputResult` and I would like some feedback for a better solution. Can you call `from_status()` unconditionally and `match`...

@zarik5 This patch simply fails to explain why the caller can't match on the `Err` results from these functions, it seems there's nothing wrong with the current implementation hence this...

Any documentation we can quote that states that these functions return both `AMEDIACODEC_INFO_*` or any of the `AMEDIA_*` error values? Regardless, I'd rather see anything that doesn't return an `OutputBuffer`...

Nevertheless, don't forget to document this problem and API change in the changelog.

> In any case I fixed also a "real" bug, the `AMEDIACODEC_ERROR_INSUFFICIENT_RESOURCE` and `AMEDIACODEC_ERROR_RECLAIMED` error were recognized as a buffer index. I changed the check from "res >= 0" to...