[foreign-memaccess] RFR: RFC: move factories and helper methods to new MemorySegments class

[Maurizio Cimadamore]

On Tue, 30 Jun 2020 16:59:25 GMT, Maurizio Cimadamore <mcimadamore at openjdk.org> wrote:

> This patch explores moving several static factories and methods (as well as some instance ones such as
> fill/copy/mismatch) onto a separate "sidecar" class, namely `MemorySegments`.
> I guess overall the approach is defensible, but the API feels a bit less tight:
> 
> * having segment factories in a different class than `MemorySegment` feels odd
> * `segment.fill(42)` reads better than `MemorySegments.fill(segment, 42)`
> 
> There might be an argument that this patch is overrotating towards the static, helper class approach, but then the
> question is where to draw the line between essential and non-essential stuff.
> One possibility would be to:
> 
> * keep MemorySegment.allocateNative where it is
> * move MemorySegment.mapFromPath to MappedMemorySegment
> * move all the view-y factories, as well as array and buffer projections, and accessors on the sidecar class
> * perhaps, move the static spliterator method in the sidecar as well
> * as for copy/mismatch/fill, we could either keep them where they are now (instance methods in MemorySegment), or move
>   them onto the sidecar
> 
> Any thoughts?
> 
> Javadoc for this revision available here:
> http://cr.openjdk.java.net/~mcimadamore/panama/memaccess-segments-helpers/javadoc/jdk/incubator/foreign/package-summary.html

Closed, as we're not pursuing this strategy.

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

PR: https://git.openjdk.java.net/panama-foreign/pull/223