Re: planer picks a bad plan (seq-scan instead of index) when adding an additional join - Mailing list pgsql-general

From Thomas H.
Subject Re: planer picks a bad plan (seq-scan instead of index) when adding an additional join
Date
Msg-id 0d5701c7046d$38c35420$0201a8c0@iwing
Whole thread Raw
In response to planer picks a bad plan (seq-scan instead of index) when adding an additional join  ("Thomas H." <me@alternize.com>)
List pgsql-general
>I wrote:
>> This didn't matter a whole lot back when the planner couldn't reorder
>> outer joins, but now that it can, it's more important that the estimates
>> be accurate.
>
>> I'm not sure if this is feasible to fix before 8.2, but I'll take a
>> look.
>
> Actually, the changes were far more localized than I first feared.
> Please apply the attached patch to your copy and see what you get for
> your problem query.

thanks for the quick patch. unfortunately i'm stuck with the win32 version
for now and haven't got the possibility to compile from source (yet)...

maybe magnus can provide me a custom win32 executable of b3 that contains
this patch [and the one for the xlog lockup]?

best wishes,
thomas



pgsql-general by date:

Previous
From: Matthew Peter
Date:
Subject: Re: RULE - special variables?
Next
From: Cornelia Boenigk
Date:
Subject: Re: zero values in statistic views