Re: Improving our clauseless-join heuristics

>>I'm afraid I'm still not following you very well. Perhaps you could
>>submit a proposed patch?

Before that can you please explain in little more detail (if possible with
small example) about the idea you have told in original mail : "is there any
join clause that both these relations participate in?"

I wanted to know the detail about the idea you told to see if what I am
proposing has any merit as compare to your idea.

-----Original Message-----
From: Tom Lane [mailto:tgl(at)sss(dot)pgh(dot)pa(dot)us]
Sent: Wednesday, April 18, 2012 10:36 AM
To: Amit Kapila
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: [HACKERS] Improving our clauseless-join heuristics

Amit Kapila <amit(dot)kapila(at)huawei(dot)com> writes:
>> I might still be misunderstanding, but I think what you are suggesting
>> is that in the loop in make_rels_by_clause_joins, if we find that the
>> old_rel doesn't have a join clause/restriction with the current
>> other_rel, we check to see whether other_rel has any join clauses at
>> all, and force the join to occur anyway if it doesn't.

> It is on similar lines, but the only difference is that it will try to
join
> old_rel with other_rel list incase
> old_rel is not able to join with any of other_rel in the list with proper
> join clause between them.

I'm afraid I'm still not following you very well. Perhaps you could
submit a proposed patch?

regards, tom lane