From: | Meszaros Attila <tilla(at)draconis(dot)elte(dot)hu> |
---|---|
To: | pgsql-sql(at)hub(dot)org |
Subject: | Re: Optimizing Multiply Joins ??? |
Date: | 2000-09-14 15:46:48 |
Message-ID: | Pine.LNX.3.96.1000914151743.2542A-100000@draconis.csoma.elte.hu |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-sql |
Hi,
> Actually, as the 7.1 code currently stands, a query that uses explicit
> JOIN operators like yours does will always be implemented in exactly
> the given join order, with no searching. I haven't quite decided if
> that's a bug or a feature ...
Do you mean a "linear binary tree" like this is executed?
/\
/\ f
/\ e
/\ d
/\ c
a b
Or can we have some variations on the graph (with the same
preorder run: a,b,c,d,e,f) like this:
/\
/ \
/ /\
/\ d /\
a /\ e f
b c
We could give hints on the joins with brackets this way:
((a,(b,c)),(d,(e,f)))
I'm not sure which version of standards allows to bracket joins,
but I know sybase accepts the above form.
How difficult it looks to hack the parser to accept this form,
and pass the meaning to the planner?
Attila
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2000-09-14 16:21:47 | Re: Optimizing Multiply Joins ??? |
Previous Message | Tim Quinlan | 2000-09-14 04:57:04 | Web Hosts (off-topic) |