On Thu, Nov 19, 2015 at 05:31:32PM -0800, Steve Kargl wrote:
> On Thu, Nov 19, 2015 at 04:58:36PM -0800, Steve Kargl wrote:
> > + else
> > + {
> > + int dm;
> > +
> > + if (dim)
> > + {
> > + if (!gfc_is_constant_expr (dim))
> > + return NULL;
> > +
> > + dm = mpz_get_si (dim->value.integer);
> > + }
> > + else
> > + dm = 1;
> > +
> > + gfc_error ("Simplification of CSHIFT with an array with rank > 1 "
> > + "no yet support");
> > + }
> > +
>
> To save some time, the dim portion of the patch isn't
> correct. dim can be scalar or rank 1 array. I'll
> #if 0 ... #endif this section unless I persevere with
> the rank > 1 case.
Ugh. Too much gdb today. The above is correct. I conflated
SHIFT and DIM's requirements.
--
Steve
