Re: R-tree, order by, limit

Anton Belyaev wrote:

> Mark, thanks for the suggestion.
> I examined PostGIS some time ago. It is too complex for my simple task
> and it gives no advantages for me:

Well okay but bear in mind the PostGIS is the de-facto standard for most
open source GIS tools. Programs like QGIS et al can visualise the
content of PostGIS tables just by pointing it towards the relevant
database - the in-built PostgreSQL geometry types aren't supported by
anything as far as I know. And don't forget coordinate re-projection -
PostGIS also allows you to re-project between latitude/longitude and
local map spatial reference systems on the fly.

> For spatial indexing it uses the same GiST-based R-tree.

Not quite. The PostGIS indexes have been improved to include selectivity
functions to allow the planner to determine when it should use the
spatial index. AFAIK the in-built PostgreSQL types use fixed values, so
the choice of index usage will be incredibly naive and often wrong on
larger datasets mixing spatial and non-spatial columns as part of the
search query.

> And PostGIS does not offer that "population" or "priority" queries I need.

Maybe. But you may find the wiki at
http://postgis.refractions.net/support/wiki/ is a good starting point
for code examples.


ATB,

Mark.

--
Mark Cave-Ayland
Sirius Corporation - The Open Source Experts
http://www.siriusit.co.uk
T: +44 870 608 0063