Optimizer problems

Started by Bruce Momjianalmost 27 years ago3 messages
#1Bruce Momjian
maillist@candle.pha.pa.us

I am fixing p_ordering(now called path_order) to properly display in
outNode/readNode and friends.

This structure is never output during debug runs. Without the display,
it is hard to see if the optimizer is doing what it should.

My suspicion is that RelOptInfo is retaining many duplicate pathlist
Path's that have duplicate/useless PathOrder/key combinations, causing
optimization of large queries to be very slow.

I'll keep going.

Is it me, or have things gotten very busy in the last week on the
hackers list, with tons of new features coming in?

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026
#2Bruce Momjian
maillist@candle.pha.pa.us
In reply to: Bruce Momjian (#1)
Re: [HACKERS] Optimizer problems

Bruce Momjian wrote:

Is it me, or have things gotten very busy in the last week on the
hackers list, with tons of new features coming in?

As usual the number of FEATURES submitted is reverse
proportional to the time left until FEATURE FREEZE :-)

I think the next time we should avoid this by discussing
about to start BETA on [CORE] instead of [HACKERS] (just a
joke - don't panic folks).

What? We should announce beta from the housetops just after every
release. Who knows how many new features we could get. :-)

-- 
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026
#3Noname
jwieck@debis.com
In reply to: Bruce Momjian (#1)
Re: [HACKERS] Optimizer problems

Bruce Momjian wrote:

Is it me, or have things gotten very busy in the last week on the
hackers list, with tons of new features coming in?

As usual the number of FEATURES submitted is reverse
proportional to the time left until FEATURE FREEZE :-)

I think the next time we should avoid this by discussing
about to start BETA on [CORE] instead of [HACKERS] (just a
joke - don't panic folks).

Jan

--

#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me. #
#======================================== jwieck@debis.com (Jan Wieck) #