values types for Java

Per Bothner
Tue Oct 21 07:24:00 GMT 2003

Bryce McKinlay wrote:

> However I'm not keen on the idea of implementing this as a GCJ 
> extension with some kind of hacked-up syntax. The benefits of something 
> only supported by GCJ would be limited and the syntax would not make 
> them elegant to use.

Note I'm not suggesting any syntax changes.  I'm suggesting a
*convention* to make it easy for a compiler like GCJ to
implement a fully portable Java class as a stack-allocatable
value class instead of heap allocation.  Both the Java source
code and the Java bytecodes would be portable Java, but compiling
to native code would use a different ABI.

 > As Per suggested, the way forward would be to propose this as a JSR
 > and take it from there.

On the other hand if we had a preliminary imlementation with some
performance numbers, that might be both a selling point for gcj,
and it might make it easier to justify a JSR.

Classic route:
* A test implementation in GCJ.
* Write up the feature and the results as an conference paper.
* Try to get it accepted at a suitable conference (OOPSLA,
the usenix JVM conference are the most obvious places).
* Use the paper (even if only web-published) to justify
forming a JSR.

The JCP is very company oriented.  I don't know how difficult
it would be for an individual to sponsor a JSR, but a company
like (say) Red Hat certainly could.
	--Per Bothner

More information about the Java mailing list