To: J3                                                      07-322r1
From: Bill Long
Subject: UTIs 127 and 128 (p339): Ultimate argument of a collective
Date: 2007 November 13
References: J3/07-007r3, J3/07-270r2


Discussion

Edits are proposed to address the points made by UTI's 127 and 128.

The requirement that the same statement invoke the collective on
all images of the team (13.1) ensures that the types and kind type
parameters are the same. The requirement that the ultimate arguments
for the first co-array dummy argument shall be corresponding co-arrays
automatically implies that the amounts of data are the same. It does
not imply that the shapes, co-shapes, bounds, or length type parameters
are the same. While this is not necessary, we feel that allowing them
to differ is not a valuable freedom, so we have added a restriction to
disallow this.

We do not think there is any need to mention polymorphic objects here.

Edits to 07-007r3

[339:18-] In 13.2.4 Arguments to collective subroutines, add extra
leading sentence:

"Each actual argument to a collective subroutine shall have the same
bounds, co-bounds, and type parameters as the corresponding actual
argument on any other image of the team."

[339:19-20] In 13.2.4 Arguments to collective subroutines, change the
second sentence to:

"On any two images of the team, the ultimate arguments for the first
co-array dummy argument shall be corresponding (2.5.7) co-arrays and
the ultimate arguments of the RESULT dummy argument shall be
corresponding co-arrays."