Bug status and Stage 3

[Salvatore Filippone]

Il giorno mer, 22/10/2008 alle 14.59 +0200, Paul Richard Thomas ha
scritto:
> Salvatore,
> 
> 
> > If my user perspective is of any help, I would say that jumping onto
> > this part of the Fortran 2008  standard BEFORE having the bulk of
> > Fortran 2003 done, makes no sense.
> 
> I agree with you totally.
> 
> > Indeed I am rather surprised that the committee included substantial new
> > features at a time when no fully compliant F2003 compiler is available
> > (IIRC we now have XLF 12 essentially complete, but it's only been out
> > for what? A couple months?); had it been me, I would only have
> > sanctioned submodules (which where a TR and an extension of concepts
> > already in).
> 
> I guess that co-arrays could be said to be in the same class - a TR
> exists and several vendors have supplied implementations for a while.
Hm. Well, I disagree (slightly), because submodules extend and fix a
very basic feature of Fortran 90: the use of MODULEs can be much more
effective if PRIVATE entities (be it variables, type components, etc.)
can be used liberally while at the same time avoiding huge module units
and recompilation cascades, which is what submodules do in a standard
way. So in my opinion this is not a new thing, it is a fix for a bug in
the previous standards (90, 95 and 2003), a bug affecting a universally
supported feature of the language, i.e. modules; whereas co-arrays are a
completely separate (and in some sense orthogonal) thing. 

> An immediate trigger for us is the implementation of co-arrays by g95.
Which is however quite debatable, as it imposes licensing requirements
for running on more than 5 images (whatever that means, but I take that
a run on 128 processors would break that limit); this sort of negates
the whole idea of a free compiler for parallel computing.

Anyway, these are second-order comments, looks like we agree that the
F2003 OOP stuff ought to be finished first. 

Cheers
Salvatore