I had your prior message on my "to look at" list; the behavior seems
pretty odd. Two questions:
1. Which version are you running, exactly?
2. Might 'peripherals & access' happen to be the largest (last in
sort order) keyword present in your table?
If #2 is true, then this might represent some sort of boundary-condition
misbehavior in the cost estimator. But I don't see anything like it
happening in current sources.
regards, tom lane
In response to
- bug in index? at 2000-07-07 18:56:07 from Martin, Sylvain R. (LNG)
pgsql-bugs by date
|Next:||From: ryan||Date: 2000-07-07 21:30:03|
|Subject: "New" bug?? Serious - crashes backend.|
|Previous:||From: Bruce Momjian||Date: 2000-07-07 19:13:05|
|Subject: Re: ps_status.h on FreeBSD 4.0 problems and fix|