Re: [boost] Proposed Geometry Library

From: Barend Gehrels (barend_at_[hidden])
Date: 2008-05-20 10:45:33

Hi Tom,

Thanks for your interest, sorry for my late reply.

Tom Brinkman wrote:
> I've been looking at the proposed boost geometry library at:
> http://geometrylibrary.geodan.nl/geometry.html
>
> Two issues so far:
>
> 1) geometry::linestring seems unecessary, it just wraps vector, deque,
> list.  Why?  If every library wrapped the standard containers in this
> way, we'd have an explosion of these wrappers.  The library should
> just use vector, deque, etc. directly.
>   

You are right. The first preview of the library used the linestring (and 
the linear_ring, which is similar) directly. Based on comments on the 
list on this, many of these occurances were replaced by iterator pairs.
So now the linestring probably might go completely.

Until now I didn't do this, the library is built following the 
conventions of the OGC (open geospatial consortium). Basic OGC 
geometries are point, linestring, polygon and multi_point, 
multi_linestring and multi_polygon, where a polygon consists of 
linear_ring's. If we omit one or many of those geometries, it would be 
less clear in that perspective.

It is actually more a  typedef then a wrapper, it could be a future 
template typedef.

> 2) The library should support std::back_inserter.  Functions, such as
> geometry::simplify should work like this:
>
> std::vector<geometry::point_xy<int> > points;
> boost::geometry::simplify(oldpoints, std::back_inserter(points));
>
>   
I agree, I was not aware that this one indeed doesn't accept a 
back_inserter. Other functions such as from_wkt support back_inserter 
and the intention is that this is possible at all places, we will work 
on this  for the next version.

Regards, Barend