Re: [boost] [geometry] area, length, centroid, behavior

Subject: Re: [boost] [geometry] area, length, centroid, behavior
From: Barend Gehrels (barend_at_[hidden])
Date: 2009-03-07 05:24:21

Hi Luke,
> I'm still confused. Why isn't scanline suitable for floating point? 
> Isn't generalize line intersection with floating point generally 
> implemented as a scanline patterned after Bentley-Ottmann?
OK, you mean the sweep line. Yes, that is suitable for floating point, 
sure. So we agree.

> If your union algorithm doesn't work by scanline and doesn't work by 
> rule-based graph traversal how does it work? These are the only two 
> methods I've heard of.
It uses Weiler Atherton (http://tinyurl.com/ajnqso), but right, that is 
a form of graph traversal (but they don't call it that way). So here 
also we agree.

Barend*
*