[vectorIntrinsics] RFR: Load/store short vectors from/to char[]

[Paul Sandoz]

On Mon, 16 Nov 2020 22:18:33 GMT, Ludovic Henry <luhenry at openjdk.org> wrote:

>> I was hoping to avoid adding support for `CharBuffer` load/store. We have so far avoided adding support for `IntBuffer` etc. When `MemorySegment` exits incubation we may even remove the `ByteBuffer` methods (since segment can wrap them).
>> 
>> Can you operate on `char[]` instead for now? I see your code requires that the `dst` buffer has a backing `char[]` array, although perhaps you intend to support direct `CharBuffer` too?
>
>> Can you operate on char[] instead for now? I see your code requires that the dst buffer has a backing char[] array, although perhaps you intend to support direct CharBuffer too?
> 
> I absolutely can only rely on a CharBuffer backed by an array only. Adding support for CharBuffer without an array is not required to make it work and to use it as an experiment, it would just make it more feature complete. I was just raising the question as I wasn't sure if it was an oversight or not, and from your explanation, it isn't.

Unsure, we might if the algorithms demand that?

-------------

PR: https://git.openjdk.java.net/panama-vector/pull/22