[postgis-devel] Xing Lin's SoC project (raster support)

[Tim Keitt]

On 7/12/07, Xing Lin <solo.lin at gmail.com> wrote:
> Sigh!!!!!!
>
> No any imitation is accpetable ? I am not a fan of Oracle, but here when I
> read his documentation on GeoRaster, it meets my ideas very much when I
> thought about this raster model. I think it is a good model. Why don't they
> allow us to take it for public ?  In fact, the Oracle GeoRaster is more or
> less the similar with ArcSDE Raster. The difference is ArcSDE provide the
> data model on the API level and oracle provide it with the database. How can
> they live together without interfering the patent?

Seems everyone is worried about IP claims these days. Yes, I wish we
could not have to worry about patents, but until the law changes,
we're stuck. A quick glance at the georaster patent gives me the
impression that it is sufficiently complex (non-obvious) to be a
problem. Of course if you worked from prior art (papers published in
the engineering/scientific literature) you would be covered even if
you overlapped the patent.

>
> Could we just take a modified version from Oracle GeoRaster or we can't take
> anything even the idea of georaster model. But it is very common in
> GeoScience and how could we avoid using it?

My understanding is that the patent covers the method of
implementation, not the idea. You can use the same idea, but the
implementation cannot be the same as described in the patent. I still
agree though that the patent is probably overly broad and covers
things that have been done many times in academia and elsewhere. The
problem is that even if it is invalid owing to prior art, the cost of
defending against a claim is prohibitive.

THK

>
> Tim, your way might be possible for the time being if we can't not take the
> previous one. I will look into the GDAL and PostGIS 0.1 soon.
>
> Thanks.
>
> Xing
>
>
>
> On 7/12/07, Tim Keitt <tkeitt at gmail.com> wrote:
>
>
> > Seems the shortest path is to add a single type to postgis that
> > represents a raster tile (if I remember, something like that may
> > already exist). Tiles would act as square polygons from postgis'
> > perspective ie would use the same gist indexing and so on. Existing
> > functions like intersect would act on the polygon bounding box. A
> > loader/dumper could easily be put together based on GDAL.
> >
> > For raster/vector ops, perhaps the simplest thing (except maybe doing
> > this externally) would be to explode the tiles as postgis points and
> > use existing functions. (Sounds slow, but might work and I suppose
> > some way to bundle the results back into an output tile would be
> > needed.) This would not be needed initially but would be nice to have.
> >
> > THK
> >
> > --
> > Timothy H. Keitt, University of Texas at Austin
> > Contact info and schedule at http://www.keittlab.org/tkeitt/
> > Reprints at http://www.keittlab.org/tkeitt/papers/
> > ODF attachment? See http://www.openoffice.org/
> >
>
>
>
>
>
> --
> -----------------------------------------------------------
> Xing Lin
> Geoinformatics
> KTH - Royal Institute of Technology [Kungliga Tekniska högskolan]
> SE-100 44 Stockholm
> Sweden
> -----------------------------------------------------------


-- 
Timothy H. Keitt, University of Texas at Austin
Contact info and schedule at http://www.keittlab.org/tkeitt/
Reprints at http://www.keittlab.org/tkeitt/papers/
ODF attachment? See http://www.openoffice.org/