Josh Berkus
But possible more error prone. If you crank up the default
statistics
to
50, but the index default is still 25... OTOH, you could always
have
the
setting of used for index default be whichever is greater... hmmm.
Well, I'm not 100% opposed to a multiplier. I'd like to
Scott,
I like it. Would a multiplier be acceptable?
default_stats_index_multiplier = 10
Yeah, I thought about that, but a multiplier would be harder to manage for
most people.I mean, what if your default_stats are at 25 and you want
your index_stats at 40? PITA. Also, if you
On Thu, 11 Mar 2004, Josh Berkus wrote:
Scott,
I like it. Would a multiplier be acceptable?
default_stats_index_multiplier = 10
Yeah, I thought about that, but a multiplier would be harder to manage for
most people.I mean, what if your default_stats are at 25 and you want
Scott,
But possible more error prone. If you crank up the default statistics to
50, but the index default is still 25... OTOH, you could always have the
setting of used for index default be whichever is greater... hmmm.
Well, I'm not 100% opposed to a multiplier. I'd like to take a poll
Josh Berkus [EMAIL PROTECTED] writes:
Well, I'm not 100% opposed to a multiplier. I'd like to take a poll of DBAs
to find out which they would find more accessable. But since most people
seem to be ignoring this thread, I'm not sure we'll get much response ...
Maybe you should ask on
Tom,
Maybe you should ask on -admin or -general. Personally I thought there
wasn't anything to say until someone did some experiments to show
whether an indexed-column differential is really worthwhile and what a
plausible default value would be. The idea sounds good in the abstract,
but
On Thursday 11 March 2004 14:17, Josh Berkus wrote:
Tom,
Maybe you should ask on -admin or -general. Personally I thought there
wasn't anything to say until someone did some experiments to show
whether an indexed-column differential is really worthwhile and what a
plausible default
Robert,
Do you plan on handeling primary key columns differently (since they are
likely to be unique and indexed) ?
The same as any other indexed column.
Also how will you handle column that are part of expressional indexes (where
foo is true for example) ?
See my original proposal.
Josh Berkus [EMAIL PROTECTED] writes:
Also how will you handle column that are part of expressional indexes (where
foo is true for example) ?
See my original proposal. These columns will be ignored. Expressions have
their own stats.
Yeah, I see no particular need to increase the stats
Bruce Momjian [EMAIL PROTECTED] writes:
Another idea is whether a foreign key column should get extra
statistics?
In practice, both ends of an FK relationship have to be indexed,
so I don't see that we need any extra special case for that.
regards, tom lane
Tom Lane wrote:
Bruce Momjian [EMAIL PROTECTED] writes:
Another idea is whether a foreign key column should get extra
statistics?
In practice, both ends of an FK relationship have to be indexed,
so I don't see that we need any extra special case for that.
Do they? We don't create an
Bruce,
Do they? We don't create an index automatically when using REFERENCES.
We do create an index for PRIMARY KEY.
I was just wondering if the REFERENCES column is more sensitive to join
usage and would benefit from more accurate statistics even if it doesn't
have an index.
I don't
12 matches
Mail list logo
