On 06/24/2013 06:55 AM, Tom Lane wrote:
> Andres Freund <andres@2ndquadrant.com> writes:
>> What about simply not using a keyword at that location at all? Something
>> like the attached hack?
> Generally speaking, I agree with Robert's objection. The patch in
> itself adds only one unnecessary keyword, which probably wouldn't be
> noticeable, but the argument for it implies that we should be willing
> to add a lot more equally-unnecessary keywords, which I'm not. gram.o
> is already about 10% of the entire postgres executable, which probably
> goes far towards explaining why its inner loop always shows up high in
> profiling: cache misses are routine. And the size of those tables is
> at least linear in the number of keywords --- perhaps worse than linear,
> I'm not sure. Adding a bunch of keywords *will* cost us in performance.
> I'm not willing to pay that cost for something that adds neither
> features nor spec compliance.
Where are we with this patch? Fabien, are you going to submit an
updated version which addresses the objections, or should I mark it
Returned With Feedback?
--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com