Re: GNAT compiler switches and optimization

[Warner BRUNS <Warner.Bruns@cern.ch>]

Damien Carbonne wrote:
> Jeffrey Creem a �crit :
>> Followup on the bug report.
>>
>> One of the comments asserted that the two programs were not equivilent 
>> though I am not yet 100% convinced that I believe it yet.
>>
> <snip>
>>
>>
>> Can someone that understands FORTRAN better make an argument about the 
>> "closeness" of this approach v.s. the other?
> 
> I've been following this thread for some times and I'm not a Fortran 
> Guru,however, IIRC, Fortran arrays and Ada arrays don't have the same 
> memory layout (Something like row major order and column major order).
> I've not compared programs of this thread with care, but I've the 
> feeling that this could change things.
> Some years ago, I had to mix Ada and Fortran, and I remember that we had 
> to invert loops order to obtain the same result.
> One would need to add 'pragma Convention (Fortran, Real_Matrix)' on Ada 
> side to obtain the same result, or to exchange loops.
> My knowledge of Fortran was limited to Fortran 77, and I don't know if 
> this is still valid with Fortran 95 or later.
> But if is still true, I would not consider the 2 programs as equivalent.
> 
> Regards
> 
> Damien Carbonne
> 
A problem with Ada is that the memory layout of multidimensional arrays
is not specified in the standard. If you want to have the same memory
layout for all Ada compilers, you have to specify that a multimensional
array should follow the convention of a foreign language, eg. Fortran.
    TYPE DefReal IS DIGITS 12;
    TYPE DefReal2D IS ARRAY(Integer RANGE <>,
                            Integer RANGE <>) OF DefReal;
    PRAGMA CONVENTION (Fortran, DefReal2D);

There is something missing in the Ada standard.

Warner