(j3.2006) Assumed type

[Malcolm Cohen]

>Did we intend to support C void* function results?

Yes, and we did in F2003.  As you note.

>  Assumed-type objects
>cannot be function results.  Is C_PTR good enough for this purpose?  If
>so, why is it not good enough for void* parameters?  I.e., why did we
>need TYPE(*)?

Strictly speaking, we did not need TYPE(*), in that we could have expanded the 
ways in which C_PTR could be used instead.  Achieving the same convenience level 
as TYPE(*) without lots of complications might have been rather difficult 
though.

Each way has pros and cons.  We decided that we thought that TYPE(*) was a 
better solution.  Maybe that was a suboptimal decision, but in hindsight it 
looks reasonable to me (at the time I favoured extended C_PTR).

Cheers,
-- 
................................Malcolm Cohen, Nihon NAG, Tokyo.