| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | andrew(at)supernews(dot)com |
| Cc: | pgsql-bugs(at)postgresql(dot)org |
| Subject: | Re: Bad mis-costing of Merge Left Join in 8.0.1 |
| Date: | 2005-04-04 01:10:15 |
| Message-ID: | 13823.1112577015@sss.pgh.pa.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-bugs |
Andrew - Supernews <andrew+nonews(at)supernews(dot)com> writes:
> The cost for the Merge Left Join is clearly preposterous, since the join
> cost can't be lower than the cost of the left branch, as it is an outer
> join and therefore that branch must be run to completion. I do not fully
> understand the cost estimation code for the merge join, but it appears to
> be reducing its total cost estimate below that of the child nodes on the
> assumption that the join can be aborted early, which is clearly not the
> case for outer joins.
Yeah, you're right ... it needs to consider whether the join is OUTER.
This bug has been there for a long time ...
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Dobrena Miteva | 2005-04-04 09:44:21 | problem with Win1251 (Cyrillic) |
| Previous Message | Andrew - Supernews | 2005-04-03 20:38:20 | Bad mis-costing of Merge Left Join in 8.0.1 |