Hi André,
On 11/10/10 9:01 PM, André Hentschel wrote:
AFAIR these failures are regression after commit 3b20cb11c8c61f141950bc5c71cfae14a35deea0. Looking at the patch it seems that these failures are strange and unexpected. Given that it worked before, we should be able to make it not without using broken() macro. Otherwise we won't be testing refcount leaks on Windows just because exactly one box has strange behavior.
Jacek
Am 11.11.2010 15:26, schrieb Jacek Caban:
True, but that's more your stuff, would need me much more time to dig in. So...patches welcome :)