(j3.2006) One more thing for the C TR

Van Snyder Van.Snyder
Mon Oct 3 20:59:24 EDT 2011


On Mon, 2011-10-03 at 06:01 -0700, Bill Long wrote:
> I think that this proposal is out of scope for the TS.  It would be more 
> appropriate as a f201x proposal, though I agree with Malcolm that there 
> are issues.

The issues seem to be very easy to address.  A compiler that offers
options to change kind numbers can provide the appropriate header files
in different subdirectories of where its distribution installer installs
things.  One can then use command-line options to select which directory
to search.  The onus is on the developer of the "make" script (or
whatever one uses) to select command-line options consistently.

> 
> Cheers,
> Bill
> 
> 
> On 10/2/11 10:07 PM, Malcolm Cohen wrote:
> >> It would be useful if the C TR required a header file containing the
> >> values of the named constants in Table 15.2 in the 2008 standard, with
> >> the names specified there, maybe in upper case.
> >
> > I disagree.  If the Fortran compiler has options that change the kind numbering,
> > this would be the opposite of useful.
> >
> >> I needed to test whether c_int == kind(0) (alternatively c_long) and
> >
> > In F2008 you can have pass the storage size and check that.
> >
> > Cheers,
> 




More information about the J3 mailing list