[EMAIL PROTECTED] wrote on 01/25/2007 12:29:13
PM:

> Bryce L Nordgren wrote:
> > Yes, this is the JTS wrapper approach (code from SYS Technologies).
> Cool - has Sanjay looked at this code? Can we grab it into
> unsupported/geometry module (and get finished faster?)

I think Sanjay examined this code at the outset.  His code is optimized
differently than the JTS wrappers.  I think the two implementations should
be complementary when they are done.  Users should be able to use whichever
implementation fits their problem better.  For the life of me, I can't
remember the exact difference, as these conversations took place back in
March (earlier?)

> > No, I haven't been committing recently.  I was in a truck bound for a
> > government surplus warehouse.
> >
> ? do not understand ?

If there was activity in the jts-wrapper branch recently, it wasn't me.  I
was in a truck fetching a 1200 pound satellite tracking system. :)

> Not to worry Bryce we have the Filter side isolated out on GeoTools
> 2.4,  we will need to look at Geometry to Shape however.

Hopefully we should be able to make some Shape adapters for the Geometry
GeoAPI interfaces.  This should serve to help adapt Geometry into Java2D
graphics.

But we might also want to "render" to an SVG file.  In which case, we make
SVG<->Geometry adapters.

Within five years, we may want to "render" to an OpenDocument Drawing file,
in which case we make *.odg<->Geometry adapters.

Heck, if we get into 3D (terrain rendering from a particular vantage
point), we may need to "render" to a Java3D scene graph, or a POVRay
script, or an X3D file.  Any of these would require the appropriate
geometry adapters.

With the broadest brush, it seems like each system has its own means of
representing geometric objects and each has roughly the same set of
objects.  It may be best to consider these "adapters" to be data-only
geometry implementations; including the geometry factories.  We may be able
to write large portions of the renderer (layer management, data handling)
once in an AbstractRenderer and produce any of the above renderers via
extensions which know which geometry factory to use. :)

> The Filter would turn around and call the Geometry method(s) - may be an
> interesting exercise in interoptability and the strength of the GeoAPI
> interfaces.

If we provide a "Shape" backed implementation of Geometry, Filter probably
should not assume that any of the operations work (e.g. intersects(),
contains(), etc.)  19107 does specify a series of "data-only" conformance
levels for an implementation.  Perhaps the geometry implementations should
be provided some mechanism by which they could report their conformance
level?  Filter would only work against those implementations which supply
the computational geometry implementation?

Bryce


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to