On Thursday, September 8, 2005, at 09:46 AM, Janis Johnson wrote:
In the hopes that it will help the discussion I ran regression hunts on the two test cases. The first test:struct foo { friend class bar; void screw (bar&); }; is rejected starting with this 4.0 patch: http://gcc.gnu.org/ml/gcc-cvs/2005-05/msg01007.html
Yeah, we already hit that one, we had to remove it from our tree. :-( Another fine example of the type of thing we just don't want at times in a x.y.n release when n != 0. I appreciate it adds a new feature that no existing code uses, but, that doesn't come for free.
