transparent lambda

Neal Gafter neal at gafter.com
Mon Jan 4 10:04:41 PST 2010


On Mon, Jan 4, 2010 at 5:14 AM, Peter Levart <peter.levart at marand.si> wrote:

> I checked the "Jump" and "UnmatchedTransfer" classes in the BGGA 2008-08-11
> prototype on javac.info. The abstract Jump class has an abstract thread()
> method that returns a thread. I haven't looked into what code is generated
> by the prototype javac in such cases, but I suspect javac generates Jump
> sub-classes for particular non-local transfers. Do you perhaps know if there
> is a document describing the implementation details of the prototype
> regarding non-local transfers?
>
> I may have unknowingly re-spawn the same idea once again, but does the BGGA
> prototype generate code with same semantics as described by Neal when he
> wrote the following:
>
>
> http://mail.openjdk.java.net/pipermail/closures-dev/2009-December/000506.html
>

BGGA does approximately what I described a transparent lambda would do.  The
scheme for the generated code is described here <
http://mail.openjdk.java.net/pipermail/closures-dev/2008-July/000154.html>
and here <
http://mail.openjdk.java.net/pipermail/closures-dev/2008-July/000175.html>.

I thought he was looking for a solution of how the desired semantics could
> be implemented.
>

No.  BGGA is already implemented.  What this thread is looking for is a
specification (syntax) that provides both transparent lambda expressions and
avoids this class of puzzle.  CfJ 0.6 a/b is my suggestion for how to do it.

Cheers,
Neal
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/closures-dev/attachments/20100104/88166280/attachment.html 


More information about the closures-dev mailing list