On Tue, 6 Dec 2005, Heikki Toivonen wrote:
Andi Vajda wrote:
I think it's still too early in the Chandler release cycle to run the
Since some others echoed this it seems clear we will go with the case by
case method.
improvement. For instance, most of the performance tests used in 0.6 are
too far from end user 'reality' to make regressions an auto-backed out
showstopper. A goal for 0.7 in the performance tracking area should be
to have more tests reflecting real end-user situations.
That's a little vague. What would you change, explicitly?
I was thinking about all the perf tests that start from an empty repository.
If our end users were to always start from an empty repository, we might as
well eliminate the repository.
Andi..
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Open Source Applications Foundation "Dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/dev
