Hi,
this is a rather nasty 4.6/4.7 Regression, ice-on-valid, and the fix
seems simple. Anyway, tested (for real ;) on x86_64-linux.
Ok mainline and branch?
Thanks,
Paolo.
/////////////////
/cp
2011-10-28 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/50901
* call.c (build_new_op_1): Handle ABS_EXPR together with the
other unary EXPR.
/testsuite
2011-10-28 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/50901
* g++.dg/cpp0x/pr50901.C: New.
Index: testsuite/g++.dg/cpp0x/pr50901.C
===================================================================
--- testsuite/g++.dg/cpp0x/pr50901.C (revision 0)
+++ testsuite/g++.dg/cpp0x/pr50901.C (revision 0)
@@ -0,0 +1,9 @@
+// { dg-options "-std=c++0x" }
+
+template<class T> int foo(int a)
+{
+ const unsigned b = a < 0 ? -a : a;
+ return 0;
+}
+
+int i = foo<float>(1);
Index: cp/call.c
===================================================================
--- cp/call.c (revision 180626)
+++ cp/call.c (working copy)
@@ -5223,6 +5223,7 @@ build_new_op_1 (enum tree_code code, int flags, tr
case POSTDECREMENT_EXPR:
case REALPART_EXPR:
case IMAGPART_EXPR:
+ case ABS_EXPR:
return cp_build_unary_op (code, arg1, candidates != 0, complain);
case ARRAY_REF:
