Comments on: avoid unnecessary updates

By: Laurent Schneider

Laurent Schneider — Mon, 09 Jul 2007 08:00:38 +0000

Pawel,
I agree with 1 and probably 3. I have not found a case where decode would be slower, but it may exists.

I am not convinced about maintanability. Decode is a trick after all

By: Sokrates

Sokrates — Mon, 09 Jul 2007 07:55:43 +0000

I don’t assume full table scan situation

By: Paweł Barut

Paweł Barut — Sun, 08 Jul 2007 09:56:03 +0000

In my opinion decode is better because:
1. it’s shorter
2. it’s easier to maintain ‘and’ & ‘or’ conditions in complex queries
3. it runs at the same speed (assuming full table scan situation)

By: Adrian Smith

Adrian Smith — Sat, 07 Jul 2007 11:02:22 +0000

Not only is the query faster (as it updates fewer rows), but it also places write-locks on fewer rows. This is normally a good thing too!

By: Laurent Schneider

Laurent Schneider — Sat, 07 Jul 2007 05:06:48 +0000

yes decode is possible too, but is it better? Well at least it is faster to type…

By: Paweł Barut

Paweł Barut — Fri, 06 Jul 2007 18:16:12 +0000

Laurent,

I would rather use this:

update t set x=:a
where decode(x,:a,’Y',’N') = ‘N’;

Paweł