[Cosmo-dev] Additional performance tests

[Jared Rhine]

Brian Moseley wrote:
> what exactly happens in the
> "kiloprop.caldata.mitch-3.1300N.delete-coll" test? weird to see that
> deletion is so slow.

All the "delete-coll" operations are the same invocation of kiloprop with
the delete operation (DAV.pm-based I think).  So the delete-coll operation
for 10b-500N.delete-coll is the same as the mitch-3.1300N.

See the series of "run" commands in:

  http://svn.osafoundation.org/sandbox/osaf.us/bin/run-propfind-test-simple

for the exact sequence; I run a few tests, the delete the collection, then
run some more, then delete the collection again, etc.  You can also
follow-along in the linked run outputs by searching for the "delete-coll"
string in your browser.

If you can think some variant slices (number of bytes? number of resources?
number of properties per resource? past history of the collection? order of
operations? etc) it shouldn't be hard to try some variants using the same
codebase.

-- Jared