200.sixtrack is miscomparing on x86_64 and i686 (-O2) since Mar 19th

[Dominique Dhumieres]

> That is not the GCC development process.  If someone confirms that the
> patch caused the bug, the 48 hour clock starts.  If the bug is not fixed,
> the patch can be reverted.

It seems to me absolutely unfair to revert a patch fixing a bug in an open
test case because it breaks something in a hidden code the mainteners have 
no access to.  If the SPEC guys want the problem fixed they have to either 
send the code to (selected) maintainers (under NDA if they want) or to 
reduce the problem to a test they can expose in the open.

Dominique