(This discussion originally arose on the FMaps list, but I have taken the
liberty of cc:ing the OSVecDB list. I would also note that Franck Martin
is actively implementing parts of the OGC Simple Features "Types" specification
for PostgreSQL).
Felipe G. Nievinski wrote:
>Maybe a talk with Frank Warmerdam <warmerda@...> would be good. He is
>heading the working group that is developing the OSVecDB (see the email
>attached describing it).
Felipe,
I am afraid my efforts on OSVecDB have flagged somewhat; however, I am also watching
(and learning from) the FMaps project.
Franck Martin wrote:
> Frank Warmerdam is on this list, so I'm awaiting his comments on the structure and his proposed structure.
>
> I think Frank, is trying to build a library for PG that will follow the OpenGIS spec very closely.
> Our goals are very similar, but I have some problems with
> OpenGIS. I hope that our devs will help each other, by sharing code.
I agree with Franck that our approach differs somewhat, but I hope to achieve
some level of sharing. In terms of actual code output in the short term, it is
my expectation to build tools around storing geometry in a variety of databases
in Well Known Binary format, with no "type" extensions to the database. Franck
is looking for more support from the database by implementing the "geoobj" and is
willing to limited himself to PostgreSQL for now. Both approaches are reasonable,
and I hope in some broad sense "OSVecDB" can result in approaches that makes it
easy for applications to work off either type of database.
However, I would like to comment briefly on a few of the points that Franck raises:
> My problems (in short):
> -OpenGIS is 2D vector
There is a 2.5D extension proposed (I am unclear on whether it was even adopted),
and I am intending to use it. 2.5D implies a Z is kept for each vertex, but that
the geometries are still fundamentally 2D surfaces with a 3D component, rather than
real 3D geometries in a CAD sense.
> -OpenGIS uses Well known Projection Systems which are VERY limited in definition
The OpenGIS "Well Known Text" format can ecompass a wide variety of projections.
However, definitions for things like vertical datums is still a bit shakey. Various
work is ongoing to improve this. I very strongly feel that "Well Known Text" format,
or an XML incarnation of it, can prove very important even beyond the narrow scope
of strict OpenGIS Simple Features implementations, as a mechanism for interchange of
coordinate system definitions. It is my hope to provide open source libraries to
support this, and to a modest degree this already exists in the form of the
OGRSpatialReference class (for parsing and forming well known text strings) and it's
binding on top of PROJ.4.
> -OpenGIS uses a schema which does not support metadata as drafted by ISO, ANZLIC, FGDC and others...
This is true. There doesn't seem to be a clear approach to transporting
metadata in the simple features standard, though I think there are attempts
to address this in various catalog and services standards from OGC that I haven't
reviewed in detail.
> -I have problems to understand database schemas, so I made mine...
> -OpenGIS has no extension for Raster data
There is a specification for this, which I believe is in the "revision working group"
stage prior to format adoption, likely within the next few months. However, it is
primarily an API specification, and I don't think the storage of raster data in a
database has been addressed.
> -I'm too slow on this project
> -I think it is time that the OpenSource community shows who is making the standards
>(I'm megalo isn't it?)
I prefer to have the open source community take a leadership role in implementing
and promoting existing standards, but if existing standards are clearly inadequate
we should be prepared to go beyond them. I do think the open source can (and almost
has to) play a major role in providing reference libraries and implementations of
standards if they are to be successful. For instance, I think the GeoTIFF effort
was an example of an informal industry effort to come up with a standard that was
largely successful because of the public domain libraries supporting the standard
written by Niles Ritter.
Best regards,
---------------------------------------+-----------------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam@...
light and sound - activate the windows | http://pobox.com/~warmerda
and watch the world go round - Rush | Geospatial Programmer for Rent
|