Test Case for stats drop post drop/add partitions?

I am losing stats post- major partition DDL operations on 817400. It's bug 2306067.
I did the research, asked for a one-off from ML, and the tech said 'we need proof'.

I did the work on my test area, but instead of dropping 45-odd partitions and
adding 30-odd, I only did one. And sure enough: it didn't affect the stats (tab or part; global or otherwise).

Now, I really don't feel like 'pretending' to do a multi=terabytes maintenance
to feed somebody's rat at oracle support, nor would my DEV and QA department appreciate my torching their only test data, but I recognize the need to provide
justification for the one-off patch.

Yet, I am getting tired of force-feeding the db stats after every partition mgmt operation. I'd like to patch the bug.

So, the question(s):

1. Has anyone run into this (8174/Solaris 8, althought it's generic)
2. Has anyone got a reasonably provocative test harness for this they'd care to share?

Basically, I just want to show HAL-9000 at Oracle Support enough information to get the one-off and supporting information.

TIA, Ross

--
http://www.freelists.org/webpage/oracle-l