Geometry Boost List: [ggl] Considering GEOS

Subject: [ggl] Considering GEOS
From: Mateusz Loskot (mateusz)
Date: 2009-04-15 16:33:16

Barend Gehrels wrote:
> Hi Mateusz,
> 
>> When I found GGL and started to work with it, I got an idea of 
>> porting some of GEOS [1] algorithms to GGL.
> 
> Actually I had seen that and had to think about it. Personally I 
> don't think it is really useful to us. Our approach is totally 
> different from theirs.

Yes, I'm aware of that.

> In the past I've looked to some implementations (not GEOS but e.g. 
> geometryalgorithms). It is possible but in the end they turn out to 
> be completely rewritten. It is justed an "influence" like reading an 
> article.

I've been thinking of looking at some solutions,
constructions, algorithms and re-implement them completely,
according to GGL approach.

> Besides that I'm afraid that if we "copy" GEOS code we'll also get (a
>  part of) its performance.

It is possible.

> Finally there is the license issue.

Right.

> It might seem attractive because they have much more (= all OGC) 
> operations implemented. However, if we continue the step of the new 
> segment intersection, together with the (completely revised) steps in
>  the polygon clip (Weiler Atherton graph traversal), we have 
> touch,overlaps,intersects,relate,disjoint,equals,intersection,union 
> and buffer all implemented so it then becomes really useful.

Absolutely, I can see this potential.

> More about this will follow.

OK, no rush.

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net
Charter Member of OSGeo, http://osgeo.org