Alban Hertroys <[EMAIL PROTECTED]> writes:
> Mm_object is always larger than any other table in the database, as
> every table joins with (different) records in it to determine it's otype
> and owner. So I don't understand how a fraction of any of those tables
> could be larger than mm_object as
Tom Lane wrote:
Alban Hertroys <[EMAIL PROTECTED]> writes:
I'm confused too. Would it be possible for you to send me a dump of
your database?
Attached is a cleaned out database, the full schema is included, but
only the relevant tables contain any data.
Thanks. After digging through it a
Alvaro Herrera <[EMAIL PROTECTED]> writes:
> Tom Lane wrote:
>> We might be able to do something about actually solving the statistical
>> problem in 8.3, but I fear it's too late to think about it for 8.2.
> I take it you mean you already have a very concrete idea on how to solve
> it. Come on,
Tom Lane wrote:
> We might be able to do something about actually solving the statistical
> problem in 8.3, but I fear it's too late to think about it for 8.2.
I take it you mean you already have a very concrete idea on how to solve
it. Come on, illuminate us poor dumb souls.
--
Alvaro Herrera
I wrote:
> Thanks. After digging through it a bit, I understand what's happening,
> but I'm not seeing any simple fix.
I forgot to mention that although I could reproduce your bad plan in
8.1, CVS HEAD doesn't fall into the trap. I don't believe we've done
anything to fix the fundamental problem
Alban Hertroys <[EMAIL PROTECTED]> writes:
>> I'm confused too. Would it be possible for you to send me a dump of
>> your database?
> Attached is a cleaned out database, the full schema is included, but
> only the relevant tables contain any data.
Thanks. After digging through it a bit, I unde
