Robert Bragg
Robert Bragg
I realized I did still have a JRE in my PATH for AndroidStudio but just re-checking after removing that it's still working ok for me. On Windows 10 ver: 21h2...
It sounds pretty surprising that having a relative path is what makes a difference for triggering a crash. That makes it seem like the DLL search order is somehow finding...
> I believe open terminals cache PATH variables I explicitly set the PATH in the shell so this wouldn't have affected testing in my case. To be sure I also...
At least one possibly relevant difference does seem to be that you're testing on a somewhat older build of Windows 10 from May 2020 vs November 2021.
oh, curious. I just tried running the binary directly (after also adding a `println("All OK");` to the end of `main()` and that ran ok for me: ``` $ ../target/debug/examples/relative_jvm.exe All...
(that's still with a bare-bones PATH and no JAVA_* variables set either)
Me neither atm - it sounds pretty darn strange so I'm not really sure what `jni-rs` could automatically do to help - while I still don't really have a good...
Just to connect the dots; this also relates to an older issue: https://github.com/jni-rs/jni-rs/issues/187
> Note that JavaVM::get_env isn't really safe either, and you can use it to create local references with invalid lifetimes, same as JNIEnv::unsafe_clone. I'm of the opinion that it should...
Cool, I'd like to make a jni release soon but if I'm honest I'm not currently expecting it to happen before the 0.5 release of android-activity or winit 0.29.