Vijay Saraswat <vi...@saraswat.org> wrote on 04/11/2013 05:19:14 AM:
>
> On 4/11/13 3:43 AM, Andreas Zwinkau wrote:
> > Does X10 specify a memory model?
> >
> > I found one related publication [0], but the language report does not
> > define anything. Is X10 supposed to inherit the Java memory model? If
> > yes, then is this really preserved with the C++ backend?
> >
> >
> We have not specified one formally.
>
> Note that finish async programs define a natural HB -- all memory
> actions of asyncs governed by the finish happen before memory actions
> that happen after the finish.
>
> Dave (G) -- when you have time can you summarize the implemented rules?
The implementation does not do anything special about the memory model
(since it isn't specified).
We make no attempt to match Java memory model on Native X10 (C++ backend).
Also, even with Managed X10 (Java backend), you get a slightly weaker
memory model than you might have expected because final fields at the X10
level are not final fields in the generated Java code (due to mismatches
between X10 and Java constructor semantics, which prevent us from doing a
direct mapping from X10 constructors to Java constructors).
--dave
------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
X10-users mailing list
X10-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/x10-users
