Ole Solberg wrote:
Nothing special for this machine compared to other machines used in the
daily tests. E.g. the "SunOS-5.10 i86pc-i386" machine in the daily test
is run on a machine which is identical and this has not shown the problem.
I did a run with the "TinderBox_Derby 418131" on a supposedly identical
machine and did *not* see the error there.
The "tinderbox" test machine seems to be the guilty.
Guilty of what, though, I'm not sure...I wonder if this issue might have
anything in common with the diffs that occur in wisconsin, namely DERBY-937? In
fact, that's one of the reasons I brought it up--I was wondering if something
changed on the machine to cause this diff, and if so then I was hoping that 1)
whatever changed might give us some insight into what's causing the optimizer to
choose a different plan, and 2) we could track the wisconsin diffs down to the
same problem. But it looks like nothing changed on the machine and the
wisconsin diff has had some developments recently, so perhaps this is unrelated...
I will move the tinderbox test to a different machine to hopefully avoid
the noise we have now.
Thanks Ole,
Army
