[postgis-users] Large point table

[Shane Spencer]

stupid mouse..

As I was saying....

You would be better off pre-transforming them with ogr2ogr.. dump them to a
shapefile, write a quicky script to move the points around to match your
grid (if that is what you are indeed doing) then insert them into the
database as the right projection.

It sounds like you will have a lot of duplicate points.  I recommend you
only insert distinct points into a table, then make a reference table with
smaller data types and a primary key referencing the point table.  Python
could handle that kind of insert situation pretty easily.

On 9/25/07, Shane Spencer <shane at bogomip.com> wrote:
>
> you would be better off pre-transforming them with ogr2ogr.. dump them to
> a shapefile, write a q
>
> On 9/25/07, Stephen Crawford < src176 at psu.edu> wrote:
> >
> > All,
> >
> > I have a table of about 30 million point observations.  The geometries
> > are
> > stored in lon/lat., with a gist index on the column.  For display and
> > analysis I want to use an albers projection....mostly I'm aggregating
> > the
> > points to a grid.  Currently I just do the transform(geom) in my
> > queries.
> > Would it be better for me to perhaps create a view with the
> > tranformation
> > instead of in my query....or should I instead create another geometry
> > column
> > in the original table, with the tranformation and do another gist index
> > on
> > the new column?  What's the best approach?
> >
> > Thanks,
> > Steve
> >
> >
> > Stephen Crawford
> > Center for Environmental Informatics
> > The Pennsylvania State University
> >
> > _______________________________________________
> > postgis-users mailing list
> > postgis-users at postgis.refractions.net
> > http://postgis.refractions.net/mailman/listinfo/postgis-users
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20070925/05230235/attachment.html>