Re: [Hibernate-devel] what to do about select distinct.....

[Gavin_King/Cirrus%<CI...@ci...>]

Okay, there seems general agreement on this. So last night I actually tried
the change out. Two observations:


(1) Only three of my tests failed after the change + I have some fairly
wierd looking queries in there. I added "distinct" to the three offending
queries + all was working again.
(2) I think I spotted a measurable performance improvement. I need to do
some more tests to comfirm this.


So I checked that change in.


Incidentally benchmarking Hibernate has been a difficult problem because
its essentially such a small overhead above the actual JDBC calls (now
generally about 10% on my machines) that comparisons get flooded out by
random variation in the JDBC processing.