"John D. Burger" <[EMAIL PROTECTED]> writes:
> Tom Lane replied:
>> But the alternative is probably even worse: without that
>> allwordrelations has to be joined to w1 and w2 simultaneously, meaning
>> that the unconstrained cartesian product of w1 and w2 has to be formed
>> first.
> Hmm, but woul
Tom Lane replied:
I have two queries for looking up related words which I think should
be equivalent, but 7.4.8 comes up with very different plans.
They're not at all equivalent:
If there are duplicate word1id,word2id entries in allwordrelations,
the
first query will produce duplicate out
"John D. Burger" <[EMAIL PROTECTED]> writes:
> I have two queries for looking up related words which I think should
> be equivalent, but 7.4.8 comes up with very different plans.
They're not at all equivalent:
> explain analyze select w2.word from allwords w1 join allwordrelations
> as r on (
