On 2/15/07, Guillaume Smet <guillaume(dot)smet(at)gmail(dot)com> wrote:
> On 2/15/07, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > I think that the
> > answer is probably "because the index is lossy for this operator,
> > so it has to be checked even if the bitmap didn't become lossy".
> > You'd have to check the GIST opclass definition to be sure.
FYI I've taken a look at PostGIS source code and the index is lossy
for the operator &&:
OPERATOR 3 && RECHECK,
(for every operator in the opclass to be exact)
--
Guillaume