[Bug c++/24439] [3.4/4.0/4.1 Regression] ICE with invert conditional containing throw

[rguenth at gcc dot gnu dot org]

------- Comment #4 from rguenth at gcc dot gnu dot org  2005-10-19 10:06 -------
Like so:

Index: cp/call.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/cp/call.c,v
retrieving revision 1.560
diff -c -3 -p -r1.560 call.c
*** cp/call.c   18 Oct 2005 05:56:04 -0000      1.560
--- cp/call.c   19 Oct 2005 10:02:58 -0000
*************** build_conditional_expr (tree arg1, tree 
*** 3226,3231 ****
--- 3226,3233 ----
            arg3 = force_rvalue (arg3);
          arg3_type = TREE_TYPE (arg3);
          result_type = arg3_type;
+         arg2 = build_compound_expr (arg2, build_int_cst (result_type, 0));
+         arg2_type = result_type;
        }
        else if (TREE_CODE (arg2) != THROW_EXPR
               && TREE_CODE (arg3) == THROW_EXPR)
*************** build_conditional_expr (tree arg1, tree 
*** 3234,3239 ****
--- 3236,3243 ----
            arg2 = force_rvalue (arg2);
          arg2_type = TREE_TYPE (arg2);
          result_type = arg2_type;
+         arg3 = build_compound_expr (arg3, build_int_cst (result_type, 0));
+         arg3_type = result_type;
        }
        else if (VOID_TYPE_P (arg2_type) && VOID_TYPE_P (arg3_type))
        result_type = void_type_node;

s/build_int_cst/create_default_constructed_temporary_target_expr/

whatever that will be.  At least we don't ICE even for

struct Foo { bool operator!=(int) { return true; } bool operator!() { return
false; }};
void foo(void)
{
  int value=1;
  !(value?Foo():throw);
}

but we create funny integer constants with type Foo, which is bad.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24439