Hi Martin,

> i hadn't realized that you wanted the distance *within* the sector.

My fault - I didn't spell it out clearly in my first post.

> You
> can't use the regular Geometry.distance(Point) method for this, of course.
> I'd think about implementing a custom Sector.distance(Geometry) algorithm,
> which computes the distance via a scan of the polygon segments.

Yep, sounds good.

> One potential issue I just thought of is that the STRtree NN algorithm
> relies on the envelope-envelope distance being a "consistent proxy" with the
> actual distance between the geometry and the query geometry (the point).
> However, I think this is still the case with the Sector distance, so that
> should be fine.

Hmm... but if "consistent proxy for" means something like "has the
same rank order as" won't there be a problem ?

> Not sure about the idea of cutting up the polygons and then using
> PreparedGeometry.  Cutting them up will probably speed things up, although
> don't you then have to "reduce" the set of distances for a single polygon to
> find the minimum?

Luckily no. All I need is the minimum distance within sector from a
point to any boundary.

> And if you use the special-purpose distance and
> intersects tests I propose above, then I don't think that PreparedGeometry
> is needed?

Right - cool !  I had imagined that I'd be doing intersections as per
previous messages, hence the idea of using PreparedGeometry.

Michael

------------------------------------------------------------------------------
Sponsored by Intel(R) XDK 
Develop, test and display web and hybrid apps with a single code base.
Download it for free now!
http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
_______________________________________________
Jts-topo-suite-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jts-topo-suite-user

Reply via email to