(j3.2006) request for interpretation

[Robert Corbett]

Bill Long wrote:
> 
> Robert Corbett wrote:
> 
>>Aleksandar Donev wrote:
>>
>>  
>>
>>>Is there a compiler that actually implements what Kurt says was the intention?
>>>    
>>
>>Yes.  Intel's ifort compiler, PGI, gfortran and g95 all do.
>>  
> 
> 
> The version of PGI that I have (7.1.6) gives a compile time error.  So 
> do Pathscale and Cray.

I was working off the list given us by the user who reported the bug.
It's possible he was wrong.

> Of course, a compiler can do anything with a non-conforming program like 
> this, since there is no hard constraint involved here.  It's possible 
> that among the compilers that do allow this, it is the case of just not 
> doing the check (since it is not required), rather than a deliberate 
> decision to pick one specific over the other.
> 
> Is there any consistency in which of the two specifics in the generic 
> gets picked by the compilers that let this through?

I don't know if the compilers are consistent, but the rules for
resolving generic calls given in Section 12.4.4 do cover the case
presented in the example.  The cases Kurt W. Hirchert said he
thought were unresolved are equally unresolved under either
interpretation of 16.2.3.

Bob Corbett