This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug fortran/48720] quad precision literals do not work


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48720

--- Comment #6 from Steve Kargl <sgk at troutmask dot apl.washington.edu> 2011-04-22 15:16:02 UTC ---
On Fri, Apr 22, 2011 at 08:00:33AM -0700, Steve Kargl wrote:
> On Fri, Apr 22, 2011 at 12:47:03PM +0000, jvdelisle at frontier dot com wrote:
> > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48720
> > 
> > --- Comment #4 from jvdelisle at frontier dot com 2011-04-22 12:46:16 UTC ---
> > On 04/21/2011 09:08 PM, kargl at gcc dot gnu.org wrote:
> > --- snip ---
> > >>
> > >> I think we are not handling the 'q' as any more than an exponent marker. 'e'
> > >> works just as well.
> > >
> > > Oh my, is gfortran accepting 1.1q0.  Let's stop that
> > > right now!  Use 1.1e0_16.
> > >
> > 
> > Agree, and its in the front end somewhere.
> > 
> 
> It's match_real_constant in primary.c.
> 
> I suppose that question is do we want to disallow q.
> 

Reading the code suggests that there are other problems if we
continue to allow q.

program foo
   real(8) d
   real(16) q
   q = 1.q0_16
   d = 1.d0_8
   print *, d, q
end program foo


troutmask:sgk[203] gfc4x -o z ui.f90 
ui.f90:5.13:

   d = 1.d0_8
             1
Error: Real number at (1) has a 'd' exponent and an explicit kind

The above is required by the standard, so I think one should 
also mandate a similar error fro 1.q0_16.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]