(j3.2006) Public Comment J32034

[Van Snyder]

On Wed, 2008-07-16 at 18:09 -0500, Bill Long wrote:
> Jim Giles wrote:
> >
> > I didn't say BLOCK was a bad idea.  I didn't oppose it in my public
> > comment.  I said in this thread that allowing declarations in all other
> > block-structured constructs was a bad idea.  I still think so.
> >
> >   
> 
> I think most of J3 agrees with this sentiment.

Is the objection to declarations other than in BLOCK, or to declarations
in constructs that have several <block>s?

It's interesting that Jim finds

  CRITICAL
    BLOCK
      ...
    END BLOCK
  END CRITICAL

and

  CRITICAL
    <specification-part>
    <execution-part>
  END CRITICAL

unacceptable, but

  ASSOCIATE ...
    BLOCK
      ...
    END BLOCK
  END ASSOCIATE

should be the only allowed way to accomplish the obvious things it does.

Of course, if we want to extend the <associate-name> to allow it to have
the POINTER or ALLOCATABLE attribute, so you could allocate or
deallocate it using the <associate-name>, or give it an INTENT or TARGET
or ASYNCHRONOUS or VOLATILE or maybe even VALUE attribute, putting the
relevant declarations inside a BLOCK inside an ASSOCIATE would create a
new object.

> In addition to the arguments already made, allowing declarations on all 
> blocks would be a bunch of work for implementers for a capability that 
> is already available.

1.  Is it already available?  IIRC, we're introducing BLOCK in 2008.

2.  Is it a bunch more work to allow declarations in <block> than in
BLOCK?  I'd be surprised if this were true.

-- 
Van Snyder                    |  What fraction of Americans believe 
Van.Snyder at jpl.nasa.gov       |  Wrestling is real and NASA is fake?
Any alleged opinions are my own and have not been approved or
disapproved by JPL, CalTech, NASA, the President, or anybody else.