(j3.2006) Suggestion for Fortran 201x: Adding support for ignoring the TKR of the actual argument

[Tobias Burnus]

Tobias Burnus wrote:
>> And give both  specifics the same binding name (of the C function that will accept either).  It is sort of involved, but does not involve any modifications to the standard.
> Well, I didn't claim that it isn't possible, but creating 16 generic
> procedures ? or with two such dummy arguments 32 ? is not really convenient.

I cannot count: With two "void *"-style dummies, one either has 16 
(common case that the arguments have the same rank) or for full 
flexibility 16*16 = 256 different specific functions. By restricting one 
to the common cases, the number is lower; still, I think one should at 
least consider whether such a functionality should be added to the 
standard, allowing the same as the rather common vendor directive.

Tobias