Message text written by Robert Kernell
>Anyone have any other examples where character kind needs to be
specified?<
I noticed that none of the replies you received to this mentioned the
history of character kinds. There appeared to be a clear need, especially
in Japan, for a method to express alternative character sets in a standard
way. An early proposal was to have the NCHARACTER data type, where N stands
for Nippon and would imply Kanji characters. This is what the Japanese
delegates to WG5 pushed for strongly. Given that kinds had already been
added to the draft standard for the other data types, X3J3 had the view
that, if this were to be done at all (not all thought it worth the effort),
then it should be done in a way that was analogous to the other types, and
this was supported also by the Chinese with their various requirements.
This was what was finally agreed and adopted. It also explains why the
examples we give in "Fortran 90/95 Explained" (Sections 2.6.4 and 9.13.2)
use Kanji. Note that the Japanese delegates would probably have voted
against Fortran 90 had this feature not been adopted.
As has been pointed out, there is, after all that hoo-ha, no known example
of a compiler that supports anything other than the default kind.
In a similar vein, the Germans would probably not have voted for Fortran 90
had the varying string module not been standardized at the same time. That,
too, has found essentially zero acceptance.
It can make one wonder about the standardization process.
Regards,
Mike Metcalf
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|