Internal compiler error in egcs g77 ss 19990623

[craig@jcb-sc.com]

>There's a fair chance that the front-end is mixing up -fno-f2c and
>-fno-f2c-library somewhere.

Indeed, good guess!  I'm doing a (fresh) build with this patch in place,
so I haven't tried it yet, but expect it should work.

Also, it fixes a regression, so I'll ask Jeff to approve including it in 2.95,
once I'm sure it works.

        tq vm, (burley)

P.S. The problem should *really* be addressed by fixing the definition
of ffesymbol_is_f2c in symbol.h, but that would mean, for elegance,
cleaning up a few other references to that macro.  The rewrite will
be addressing stuff like that anyway.  In the meantime, this appears
to be the only place where this macro got used in a way that triggered
this sort of confusion.


Mon Jun 28 10:43:11 1999  Craig Burley  <craig@jcb-sc.com>

	* com.c (ffecom_prepare_expr_): A COMPLEX intrinsic needs
	a temp even if -fno-f2c.

	* version.c: Bump version.

*** g77-e/gcc/f/com.c.~1~	Wed Jun  9 13:06:43 1999
--- g77-e/gcc/f/com.c	Mon Jun 28 10:42:34 1999
*************** ffecom_prepare_expr_ (ffebld expr, ffebl
*** 13069,13073 ****
  	      s = ffebld_symter (ffebld_left (expr));
  	      if (ffesymbol_where (s) == FFEINFO_whereCONSTANT
! 		  || ! ffesymbol_is_f2c (s))
  		break;
  	    }
--- 13069,13076 ----
  	      s = ffebld_symter (ffebld_left (expr));
  	      if (ffesymbol_where (s) == FFEINFO_whereCONSTANT
! 		  || (ffesymbol_where (s) != FFEINFO_whereINTRINSIC
! 		      && ! ffesymbol_is_f2c (s))
! 		  || (ffesymbol_where (s) == FFEINFO_whereINTRINSIC
! 		      && ! ffesymbol_is_f2c_library (s)))
  		break;
  	    }
*** g77-e/gcc/f/version.c.~1~	Wed Jun  9 13:08:15 1999
--- g77-e/gcc/f/version.c	Mon Jun 28 10:43:08 1999
***************
*** 1 ****
! const char *ffe_version_string = "0.5.25 19990526 (prerelease)";
--- 1 ----
! const char *ffe_version_string = "0.5.25 19990628 (prerelease)";