(j3.2006) Are 7.2.1.2p1(1) and 7.2.1.2p1(8) correct?

[Van Snyder]

On Thu, 2015-03-12 at 13:54 +0900, Malcolm Cohen wrote:
> <<<
> 7.2.1.2p1(3) says that if the variable and expression in an intrinsic
> assignment are not conformable, the variable shall not be an allocatable
> coarray or coindexed object.  Presumably, this is so that an intrinsic
> assignment will not be an image control statement.
> 
> 7.2.1.2p1(1) and (8) only say the variable shall not be an allocatable
> coarray.  Should an allocatable coindexed object also be prohibited?
> >>>
> 
> I draw your attention to p2, which has requirements on coindexed objects.  The 
> first and last bullet points cover p1(1) and p1(8) respectively, but the middle 
> bullet point is something that is fine for a coarray but bad for a coindexed 
> object.
> 
> I don't know why p1(3) mentions coindexed objects - editorially I would have 
> thought that either all the coindexed requirements should be in p2, or they 
> should be in the relevant p1(N) items.  Since p2 middle must remain, moving the 
> p1(3) one into p2 might be appropriate.

Yeah, I read p2 and noticed it didn't cover the ground of p1(1), p1(3),
and p1(8).  I still wonder why p1(1) and p1(8) don't mention allocatable
coindexed variables.  If they appeared and got reallocated, wouldn't
that make an intrinsic assinment an image-control statement?

I don't think we want to say that assigning to an allocatable coindexed
variable is prohibited, just that assignment cannot be done if it would
result in reallocation.

> Cheers,