integer*8 parsing problem

[Steve Kargl]

On Mon, Aug 23, 2004 at 04:25:16PM -0700, Richard Henderson wrote:
> Going through a string here seems unnecessary.  Thoughts on 
> this rewrite?  Comments wrt the range of the MPZ integer vs
> the type are especially appreciated.
> 

Unless some calls to gfc_check_integer_range() are missing,
I believe we range check all integer constants against 
their corresponding int_info->huge value where int_info->huge
is set in gfc_arith_init_1().

Going through a string may be unnecessary, however, I note
that your patch is the only place in the gfortran frontend
where low level GMP details are used (ie., mp_limb_t).

-- 
Steve