Ben Alex

Any PRs that improve our Java 9 support are most welcome. Unfortunately I have limited time to work on this at the moment.

I've now had a chance to test LmdbJava with Java 9. First, the good news: * Some of our `Unsafe` use can be replaced with `com.kenai.jffi.MemoryIO` operations * Netty `Unpooled`...

As LWJGL notes, [VarHandles](http://openjdk.java.net/jeps/193) do not allow code to arbitrarily access another module's private fields. I suggest we: 1. Remove all `Unsafe` use and replace with `MemoryIO`. That means no...

Just noting here the Java 9 `module-info.java` found to work for future reference (I won't commit this given it would disrupt the build to exclude this Java file etc): ```...

> I'm not sure what this solves since lmdbjava would still be transitively dependent on Unsafe through jnrffi and Agrona The main thing it would solve is removing all illegal...

Commit 5687bd61ff8de770800fa319b4b98a0f61a92b24 shows how to add the required JVM parameters for running tests in Maven builds etc. It's implemented as a profile as we presently perform CI builds with Java...

This issue may be 5 years old, but we've been waiting for [Project Panama](https://openjdk.java.net/projects/panama/) to release long-term, stable APIs. It was only with Java 17 (14 September 2021) did this...

There are several pieces to get this working: 1. Ensuring JNR works at all with Android 2. Building the LMDB native library for Android 3. Resolving any LmdbJava-specific interop issues...

@EternalDeiwos would you be happy to try the "hello world" test I mentioned [above](https://github.com/lmdbjava/lmdbjava/issues/47#issuecomment-279077979) on Android? I do not have any Android developer toolchain or experience to move this forward...

On the speed issue, [Hash-Bench](https://github.com/benalexau/hash-bench) tests over 100 hash implementations on the JVM and finds Zero-Allocation-Hashing's XXH64 is the fastest. Feel free to check the published results for yourself for...