Hello world, another obvious fix, this time for an ice-on-invalid regression, committed after regression-testing.
Regards Thomas 2017-01-30 Thomas Koenig <tkoe...@gcc.gnu.org> PR fortran/84134 * array.c (gfc_ref_dimen_size): Whitespace fixes. If stride is zero, return false. 2017-01-30 Thomas Koenig <tkoe...@gcc.gnu.org> PR fortran/84134
Index: array.c =================================================================== --- array.c (Revision 257131) +++ array.c (Arbeitskopie) @@ -2245,9 +2245,12 @@ gfc_ref_dimen_size (gfc_array_ref *ar, int dimen, else { stride_expr = gfc_copy_expr(ar->stride[dimen]); + if(!gfc_simplify_expr(stride_expr, 1)) gfc_internal_error("Simplification error"); - if (stride_expr->expr_type != EXPR_CONSTANT) + + if (stride_expr->expr_type != EXPR_CONSTANT + || mpz_cmp_ui (stride_expr->value.integer, 0) == 0) { mpz_clear (stride); return false;
! { dg-do compile } ! PR fortran/84134 - this used to ICE. ! Test case by Gerhard Steinmetz program p integer :: i, x(3) data (x(i+1:i+2:i),i=0,1) /1,2,3/ ! { dg-error "Nonconstant array section" } end