[postgis-devel] Broken r-tree semantics in PostgreSQL (rtree_gist and rtree)

[strk at refractions.net]

On Fri, Jun 24, 2005 at 02:04:26PM +0200, strk at refractions.net wrote:
> On Fri, Jun 24, 2005 at 07:35:06AM +0100, Mark Cave-Ayland wrote:
> > For information:
> > 
> > http://archives.postgresql.org/pgsql-hackers/2005-06/msg01108.php
> 
> To summarize things up, is the problem with
> 
> 	left NOT EQUIVALENT TO ! overright
> 
> and others ?

Ok, I checked updates in pgsql and ported them to our implementation.
I've also ported use of internal_consistent always (even for leaf
nodes) as we also have a LOSSY specification (for SRID, but that is).

Regression tests work, but worked before as well (we lack a test for
this case)...

I committed the change in both head and 1.0 branches, do you think
I should port the code to 0.9 branch as well ?

--strk;