https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99765
Tobias Burnus <burnus at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Resolution|--- |INVALID
Status|NEW |RESOLVED
--- Comment #5 from Tobias Burnus <burnus at gcc dot gnu.org> ---
(In reply to Nick from comment #4)
> I see. I can't seem to find the mentioned line in f2003.
Should be there as well.
In Fortran 2008, it is in "5.2 Type declaration statements", p.88, lines 1-4
(same/similar to F2018).
In Fortran 2003, it is in "5.1.2.5 DIMENSION attribute", p. 78, ll. 3-5:
"The DIMENSION attribute specifies that an entity is an array. The rank or
rank and shape is specified by the array-spec, if there is one, in the
entity-decl, or by the array-spec in the DIMENSION attr-spec otherwise."
That choice makese sense, e.g. for:
complex(kind=my_cmplx_kind), intent(in), asynchronous, dimension(n,n) :: A, B,
C, D, v(n)
such that one does not need to repeat all the lengthy stuff just to denote the
different array spec for 'v'. (On the other hand, whether there is a need to
specify everything in several different ways and permit overriding in addition
is another question.)
Thus, the standard made a sensible choice – and the standard is the standard
:-)
> In any case, I would be fine if this is marked as invalid
Done so: CLOSE as INVALID.
> Sorry for the blurp.
Better some noise than missing some real bugs or useful improvements. Hence:
Thanks!