(j3.2006) Finalizing subroutines question

[Bill Long]

Michael Ingrassia wrote:
>> Really?  I thought it was only when the type of the variable has a final 
>> routine specified.  f03:[59:33-34].
>>     
>
> Actually, I think there is a difference between having a final routine specified
> and being finalizable.  A type is finalizable if a component has
> a final routine specified, or a component of a component has a final
> routine specified., etc. 
>   

Just for nonpointer and nonallocatable components.  But you are right 
that the situation is more general that I said above.

> If you build your coarrays completely from the part of your type forest
> that have no finalizers, you can avoid finalization completely, but using
> any one little building block (maybe a varying string with a finalizer?) 
> which needs finalization would seem to create the need to finalize
> coindexed objects.   Presumably if the little pieces need finalization
> you'd actually like that to work as designed.
>
>
>   

Not if it involves altering allocations or pointer associations on 
another image.  Jim's suggestion is probably simpler and easier to 
implement in the standard.

Cheers,
Bill

-- 
Bill Long                                   longb at cray.com
Fortran Technical Support    &              voice: 651-605-9024
Bioinformatics Software Development         fax:   651-605-9142
Cray Inc., 1340 Mendota Heights Rd., Mendota Heights, MN, 55120