db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Knut Anders Hatlen <knut.hat...@oracle.com>
Subject Re: (DERBY-5010) [patch] bad equivalence check
Date Tue, 08 Feb 2011 22:48:10 GMT
>> Knut Anders Hatlen commented on DERBY-5010:
>> -------------------------------------------
>>
>> The current code looks wrong and the suggested fix looks right.
>>
>> None of our regression tests seem to exercise this code path,
>> though. It would be good if someone could come up with a statement
>> that caused this code to be executed so that we can verify that it
>> behaves as expected after the proposed changes.

Jayaram Subramanian <rsjay1976@gmail.com> writes:

> Hi Knut,
> I am interested in taking up this task.  Let me know your views on the same.

Hi Jayaram,

Thanks for volunteering to look into this. I think a good first step
would be to find a query that actually ends up calling
BaseColumnNode.isEquivalent(). I saw from some old test coverage reports
that the method was called, but the parts of it that were touched by the
patch were never executed. For example, you can add a call to
Thread.dumpStack() in that method, run the lang test suite
(org.apache.derbyTesting.functionTests.tests.lang._Suite), and see where
it's called from.

The method was added as part of DERBY-883, so I'd expect that queries
that use it typically have aggregates and GROUP BY clauses.

-- 
Knut Anders

Mime
View raw message