[postgis-users] Broken r-tree semantics

Bill Binko bill at binko.net
Fri Jun 24 09:11:00 PDT 2005

On Fri, 24 Jun 2005 strk at refractions.net wrote:

> Hello users!
> Reference:
> http://archives.postgresql.org/pgsql-hackers/2005-06/msg01108.php
> A bug in r-tree semantics has been found and fixed.
> We ported the fix to postgis, which is also affected by
> it. Basically in some cases an index scan would fail
> catching some records that would be coutch by a
> sequential scan. 
> Now:
> - Do you think it's worth porting the patch to the 0.9 branch ?

Sounds like the query planner can choose to do an index scan without 
user's knowledge, correct?  So people who are using 0.9, but who's data 
count is low may not see the error now, but will as their tables grow?

I think if that's the case, then you need to either specifically tell 
everyone to stop using 0.9 or back-port (and I'd recommend back-porting).

> - Did anyone experienced this problem ?

I don't think so.

More information about the postgis-users mailing list