> The newer Mayfly code appears to be causing the Mifos test suite from
> trunk revision 12291 (latest as of right now) to run about 25% slower

Oh, man.  Why couldn't you have an easy problem?

> I suppose it might speed up Mayfly if it were built without debugging
> info

Well, you can try it and see if it speeds anything up, but if Java
debugging info is anything like C debugging info, it shouldn't affect
run times (the latter isn't even faulted into memory, much less looked
at, unless debugging is going on).

> Any ideas why the new version might make Mifos unit tests run slower?

Not a whole lot I can help with, unless (a) you identify a particular
Mayfly feature which slowed down (my guess is that it is just one
operation, rather than a general slowdown in all of mayfly), or (b) do
some profiling on Mayfly and find a bottleneck, or (c) figure out
which change in Mayfly caused the problem.

If you want to try (b), see
http://jkingdon2000.blogspot.com/2007/07/profiling-mayfly.html first
(including the comments, which include some important caveats).

If you want to try (a), try writing a few short performance tests
(along the lines of PerformanceTest in the mayfly sources) for each
of the statements which occur frequently in mifos's latest-schema.sql
and latest-data.sql (or whatever they are called now).

If you want to try (c), the obvious approach is straightforward but
potentially a bit tedious: binary search on mayfly versions.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

Reply via email to