Re: [boost] [Review] We're halfway through the GGL review

Subject: Re: [boost] [Review] We're halfway through the GGL review
From: Fabio Fracassi (f.fracassi_at_[hidden])
Date: 2009-11-16 12:52:49

Barend Gehrels wrote:
> Hi Fabio,
> 
>>     Hm, I tried:
>>
>>     #include <boost/ggl/ggl.hpp>
>>     #include <boost/ggl/geometries/cartesian3d.hpp>
>>     #include <boost/ggl/multi/geometries/multi_polygon.hpp>
>>     #include <boost/ggl/algorithms/union.hpp>
>>
>>     ...
>>
>>     typedef multi_polygon<polygon_3d> polygon_set_3d;
>>     polygon_set_3d ps;
>>         polygon_3d poly1;
>>     polygon_3d poly2;
>>         union_inserter(poly1, poly2, std::back_inserter(ps));
> 
> There are two things:
> 
> - I forgot one thing, sorry, the inserter-versions always have their 
> output type as a required template parameter. That one can, AFAIK, not 
> be deduced from the output iterator. And it might be a different type of 
> the input polygons. If someone knows if that is possible somehow, I 
> would be happy to know. So it should be: union_inserter<polygon_2d>(p1, 
> p2, std::back_inserter(ps1));
> 

Ok, one more argument for both an example and a union_ function then ;)

> - You're using the 3d polygons here. The union would make a 2D 
> union-operation then of your 3D polygons (so polygons with for all 
> vertices a height). It would ignore the third dimension (because it is 
> not a polyhedron union). I just tried it, to not give you again wrong 
> information. But it is not working yet like this, a.o. because the 
> functions convert, area and point-in-polygon which should support 3D 
> then. Especially area implementation would probably be different for 3D. 
> Though it is used as a helper-function in the union, so its usage could 
> be avoided.
> 

I didn't really expect 3d union to work (i.e. create a polyhedron ) I 
wanted to find out what would happen. I'm not quite sure what I would 
expect (projection on the xy plane like you suggest or another 
projection most likely).
But the docu said that 3d was not a focus of GGL yet, so its ok that 
there are missing pieces.

I'm content that the design allows for both union_(poly3d_1, poly3d_2, 
poly3d_result) and union_(poly3d_1, poly3d_2, polyhedron) to do the 
right thing, once the building blocks are in place.

Thanks and Regards

Fabio