On Wednesday 04 July 2007 23:01, Yasuki Arasaki wrote:
> Now P, with the CONTIGUOUS attribute, is associated with A(::2)
> which is not contiguous. If I read correctly, there isn't anything
> in the draft to prevent this.
There is, in fact, there is a whole messy section on the relation between
pointer association status and argument association. I suspect it is
virtually impossible to understand it unless one has some other clues (like
being on the committee when the words were changed). In particular, there are
are a few (newly added) scattered appearances of "other than an assumed-shape
dummy with the CONTIGUOUS attribute" in that section to ensure that copy
in/out is still allowed when the actual is not contiguous and the dummy is
expected to be. Pointer association status is not preserved (i.e., you get
processor-dependent or dangling pointers).
Best,
Aleks
--
Aleksandar Donev, Ph.D.
Lawrence Postdoctoral Fellow @ LLNL
High Performance Computational Materials Science and Chemistry
E-mail: [log in to unmask]
Phone: (925) 424-6816 Fax: (925) 423-0785
Address: P.O.Box 808, L-367, Livermore, CA 94551-9900
Web: http://cherrypit.princeton.edu/donev
|