Re: Bugs in planner's equivalence-class processing

On 16 October 2012 16:56, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

> Is anybody concerned about the compatibility implications of fixing this
> bug in the back branches? I'm worried about people complaining that we
> broke their application in a minor release. Maybe they were depending
> on incorrect behavior, but they might complain anyway. On the other
> hand, the fact that this hasn't been reported from the field in nine
> years suggests that not many people write queries like this.

Thanks for investigating this. My experience is that people seldom
check or understand the output of a query, they probably just figure
they didn't understand SQL and rewrite a different way, so its hard to
gauge the impact.

I think we need to see the cure before we can decide whether its worse
than the disease. And especially important is that we fix this just
once so I suggest fix and then backpatch deeply later.

This type of thing is handled in other products by having a
compatibility level, so you can decide whether you want it or not. Not
suggesting that here, yet, but its one way of mitigating the change.

--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services