[J3] C1128 in N1259

[Van Snyder]

On Sun, 2018-08-19 at 16:14 +0900, Malcolm Cohen via J3 wrote:
> One thing you are missing is the definition of finalizable:
>  
> 3.74
> finalizable
> ⟨type⟩ has a final subroutine or a nonpointer nonallocatable component
> of finalizable type
>  
> So the finalizable type subobject question does not arise, it is
> correctly prohibited by the constraint.
>  
> Maybe (he surmises) having an allocatable ultimate component ought to
> be prohibited too.
>  
> BLOCK constructs are irrelevant.

We say that variables with construct scope within DO CONCURRENT
constructs are "similar to LOCAL" in 11.1.7.5p1.  We don't say "as if,"
so C1128 doesn't apply.  Should it?

How can there be a problem for a variable declared outside a DO
CONCURRENT construct with the ALLOCATABLE attribute, whose name is used
to create a new and different one for each iteration (that's what
11.1.7.5p1 says) by the "spooky action at a distance" of a LOCAL
locality spec (which we dislike for statement function dummy arguments),
but not for one declared within a BLOCK construct within the DO
CONCURRENT construct (which is "similar to LOCAL" but not "as if"
LOCAL)?

Or a finalizable one declared in a BLOCK inside a DO CONCURRENT
construct, instead of using spooky action at a distance?

And I still don't understand why there's a problem for an allocatable
variable with LOCAL locality, but not for a variable with with LOCAL
locality that has an allocatable component.

Indeed, I don't recall why there's a problem for an allocatable or
finalizable variable with LOCAL locality, as there appears not to be a
problem for those kinds of variables declared within BLOCK constructs
within DO CONCURRENT constructs, which are "similar to LOCAL" (but not
"as if").

Those who were at Garching might remember that I had a paper about the
identity (not similarity) of LOCAL locality and variables with construct
scope within DO CONCURRENT constructs.  That paper was "no action" from
JOR, apparently without reading past the Subject line.  I had to ask
several times for it to be considered, which it finally was for ten
minutes, when we already had one foot out the door on the way to the
airport.

>  
> Cheers,
>  
> From: Van Snyder via J3 
> Sent: Sunday, August 19, 2018 2:54 PM
> To: j3 
> Cc: Van Snyder 
> Subject: [J3] C1128 in N1259
>  
> I don't understand the usefulness of most of C1128 in N1259.
> 
> It prohibits a varioable that has LOCAL locality from having the
> ALLOCATABLE attribute or being of finalizable type. It doesn't
> prohibit
> these things for subobjects of such variables. It doesn't prohibit
> these things for variables declared within BLOCK constructs within DO
> CONCURRENT constructs.
> 
> Am I missing something, or did we fail to finish this constraint, or
> is
> it overkill?
> 
> 
> 
> 
> Disclaimer
> 
> The Numerical Algorithms Group Ltd is a company registered in England
> and Wales with company number 1249803. The registered office is:
> Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.
> Please see our Privacy Notice for information on how we process
> personal data and for details of how to stop or limit communications
> from us.
> 
> This e-mail has been scanned for all viruses and malware, and may have
> been automatically archived by Mimecast Ltd, an innovator in Software
> as a Service (SaaS) for business.
> 
>