This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran 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]

Re: [Patch, fortran] PR29507 - ICE with array actual for the INDEX intrinsic.


I temporarily withdraw this patch - I had not realised how general the problem is. I started writing a "scalarizer" this lunchtime; I got as far as it compiling and then segfaulting in execution :-)

Paul

Here it is - thanks Steve and Tobias for giving me a little nudge.

Paul



:ADDPATCH fortran:

This PR is due to an ICE which occurs because the simplification
of the INDEX intrinsic has no means to handle array constructor
arguments.  The fix is a straightforward bit of gfc_expr'ionary
and hints at a more general solution in gfc_simplify_expr for all
intrinsic calls that expect a scalar argument but receive an array.
I propose to keep this PR open, even if this patch is accepted, to
function as a TODO for the more general approach. The testcase is
Steve's original.

Regested on PIV/Cygwin_NT - OK for trunk?

Paul

2006-10-19 Paul Thomas <pault@gcc.gnu.org>

	PR fortran/29507
	* simplify.c (gfc_simplify_index): If the first argument is
	an array, build an array constructor result and call this
	function for each of the elements.

2006-10-19 Paul Thomas <pault@gcc.gnu.org>

	PR fortran/29507
	* gfortran.dg/index_1.f90: New test.









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