Tom Lane wrote:
> No, not at the moment. In principle the planner could look for such
> duplicates, but it'd be wasted cycles so much of the time that I'd be
> loath to do it.
Good point - there are better places to spend time, and I imagine it'd
be an expensive thing to check too.
> There is work afoot to implement the SQL:2003 "WITH" syntax, which
> I think would offer a syntactic solution to your problem.
Yes, it would. In fact, I was thinking about the syntax seen in some
functional languages - like Haskell's `where' clause - that defines a
subexpression available to all parts of the expression. If the SQL:2003
WITH expression is anything like that it'd be very handy indeed.
In response to
pgsql-performance by date
|Next:||From: petchimuthu lingam||Date: 2008-03-13 07:21:54|
|Subject: partial index + select query performance|
|Previous:||From: Joshua D. Drake||Date: 2008-03-13 05:55:54|
|Subject: Re: Benchmark: Dell/Perc 6, 8 disk RAID 10|