2007/4/23, Tor Arntsen wrote:

> Yes, my IRIX 6.5 systems also runs the tests that the 6.2 system (with
> the 6.2 compiler) sometimes fail. I have looked at this several times
> in the past but haven't gotten past the scratch-my-head phase.
> [...]
> The way it looked to me though when I tested this was that the problem
> seemed to be random, one test could work and a rerun fail. (That would
> mean that CVS isn't to blame I guess)

Yep, so random that today's tests have succeeded.

I could be wrong, of course, but I'm more in the mood of blaming an
uninitialized variable or a buffer overflow somewhere due to its
apparently random occurrence, and since it is only happening on that
specific platform/compiler those also become suspicious.

Well, its a bit like chasing ghosts, now you see them now you don't ;-)

-- 
-=[Yang]=-