Boost logo

Geometry :

Subject: [ggl] Re: segment iterators
From: Barend Gehrels (barend)
Date: 2009-05-04 23:46:15

Hi Mateusz,

Thanks for your changes in all files again!

Bruno, for your information, we found out that using test_unit.hpp is not
necessary anymore. Mateusz removed that headerfile because the order of
inclusion turned out to be important.

> Cool.
> As you may have noticed, this is input
> iterator.
> From what I understand, it follows the initial to be able to
> adapt
> list of points, so they can be read as segments.
> Is this OK and enough? Or we need bidirectional or may
> be
> even random iterator?
> IOW, do we want to have specializations for
> other iterator categories?
I think we need random, or a version which works on segments. But for the
first version this is probably OK, the segment version will be different
anyway. Bidirectional is not necessary now but if we (ever) also support
anti-clockwise polygons we might walk through segments in reverse order.

>> We should compare the speed with
>> normal iterating.
> This could be interesting, indeed.
> First, I'm
> moving back to testing my concept of WKB IO
> then I'm going to move to the
> benchmarking thing.

Regards, Barend

Geometry list run by mateusz at