(j3.2006) Storage_size ??

[Robert Corbett]

On 07/29/13 08:29, Bill Long wrote:
> Give the answers so far, it appears that the value returned is 
> implementation-dependent, but is always loc(xx(2)) - loc(xx(1)). 

Your question exposes assumptions that need not be true.

Suppose a program defines a type as

       TYPE (T)
         REAL V(1000)
       END TYPE

and a variable of that type

       TYPE(T) :: X

Because T is not a sequence or BIND(C) type, I think it would be standard 
conforming for a processor to implement the variable X as a machine-level 
pointer.  The pointer might initially be set to NULL to signify that it is 
undefined.  If the variable first becomes defined, space would be allocated for 
the structure and it would be initialized to the proper value.  Under such a 
scheme, the value of STORAGE_SIZE(X) would be the size of a machine-level 
pointer in bits.

A similar scheme could not be used if T were a sequence type.

Bob Corbett