(j3.2006) Tobias Burnas question about namelist

Malcolm Cohen malcolm
Fri Feb 26 01:59:21 EST 2010


Van Snyder wrote:
> Tobias Burnus had a valid quibble with the words about namelist.

I agree.  I think it is rather too late to fix this other than via the interp 
process.

> Would it be enough to replace "type parameters" by "kind type
> parameters" and "shape" by "rank" at [09-007r3:111:19-20 5.6p5]?

This would seem to be the minimal edit required.

> Surely compilers can handle
> forward references by now,

I agree.  On the other hand, putting NAMELIST statements at the end of the other 
declarations is hardly a significant imposition.

> so it could be simplified to
>
> C586a (R563) A <namelist-group-object> shall be accessible in the
>      scoping unit in which the <namelist-stmt> appears.

This constraint is content-free; if the identifier for a particular entity is 
not "accessible" in a scoping unit, then in that scoping unit said identifier is 
not an identifier of that entity!  I.e. this is handled by the scoping rules 
already.

So either the minimal edit should be done, or the sentence deleted altogether.

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




More information about the J3 mailing list