On Tuesday 13 April 2004 15:17, Brian McCallister wrote: > Huh, the ODMG slower thing is interesting as Armin just reemed us all > out for letting the OTM get so much slower than ODMG ;-) Have I missed something? Are there any new performance tests? When I tested last time, OTM was somewhat faster. Of course, when you use OTM PB-style Criteria/Queries instead of ODMG-style OQLQueries, the difference would be more noticeable. But don't forget that OTM provides true ACID transactions, pessimistic and optimistic locking strategies (automatic upgrading of read lock to write lock for modified objects) and some other nice features like automatic storing of object modifications (for modified objects only), automatic creation/deletion of otm-dependent objects, long transactions support. So I guess OTM may be slower in some tests, especially when you run them against in-memory database like HSQL.
> but it is a little bit less than mature at the moment. > It is definately in the unstable category for 1.0. Completely agree. > Oleg is in the process of doing a major factoring job on parts of it to > clean it up and fix a couple bugs (otm-dependent proxied collections > not detecting deletes unless an explicit write lock is obtained on the > parent is a particularly unpleasant one for me). The refactoring is now finished, and the bug that you mentioned is now fixed, at least according to my tests. Please re-run your tests. Many thanks to you and to all who try OTM, test OTM and report bugs! ;-) Regards, Oleg --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
