[Mondrian] Mondrian Build Failed

[Julian Hyde]

> Julian wrote:
>
> Looks like a query was triggering some kind of cache 
> corruption which was causing subsequent queries to fail. I've 
> disabled the test which was causing the problem, and logged a bug.
> 
>
http://sourceforge.net/tracker/index.php?func=detail&aid=1710913&group_i
d=35302&atid=414613

Update. I fixed the problem in change 9204.

http://perforce.eigenbase.org:8080/@md=d&c=6PU@9204?ac=10 

It was nothing to do with Will's change, although Will's testcase
happened to cause the cache to be loaded with invalid data which caused
subsequent tests to fail.

The other strange thing about the bug was that it was subjective (only
certain configurations would see it) but deterministic (configurations
which saw it would see it every run). I presume that the amount of free
memory in the JVM was a major factor.

The fact that we are diagnosing bugs like this is a testament to the
breadth of the test suite we have built. It gives me confidence that the
number of correctness bugs in the mondrian code base is at an all-time
low.

Julian