db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel John Debrunner <...@debrunners.com>
Subject Re: [jira] Commented: (DERBY-649) Useful indexes not used in UNION ALL
Date Wed, 21 Dec 2005 19:12:30 GMT
Satheesh Bandaram wrote:

> Thanks for catching the regression and resolving it so quickly. Is there
> anything you cann't do? :-)

Hey, my bit was the easy bit, spot the typo in the patch given a simple
reproducible case. Had I had to actually fix the problem like you did,
it would have been a different story. :-)

> I thought the reason for having TRUE constant node was because of need
> to have the predicate in CNF form... AndNode is always at the top, which
> assists (or simplifies) logic in the optimizer. I am still not done with
> changes in 10.1 branch. I will add a comment for the need to have
> constant node.

Most likely true, actually I did mess up here by asking why there was a
'= TRUE' construct, when the real construct in the original patch was
'col = value AND TRUE'. If I had looked more carefully I hope the '='
would have jumped out at me.

Many eyes etc.

View raw message