Boost logo

Geometry :

Subject: [ggl] Problems with the difference between two polygons
From: Barend Gehrels (barend)
Date: 2011-07-21 19:25:16

Hi Enrico,

On 21-7-2011 21:40, Enrico Leoni wrote:
>> It might also be good to call boost::geometry::correct to your input
>> polygons, because they are not closed. But that does not solve your second
>> problem (maybe it solves the first).
> Using "correct" and "unique" I'm in your same situation, moreover,
> with TTMath the result is correct but with MPIR I get four polygons.

OK, thanks for the info.

> Given that I'm working on a little benchmark, I see that with TTMath
> or MPIR the execution time is noticeably higher than using only
> "double" coordinates so I ended upscaling the polygons and truncating
> the decimals (keeping the coordinates as "double"), but now I'd be
> curious to know if a higher execution time with MPIR is expected or if
> I am doing something wrong.

With TTMath the execution time is certainly higher. I don't have
experience with MPIR, based on a glance on the website it is based on
GMP so the time will also be higher. In general, I expected all high
precision libraries to be much slower.

I will look at your report again, soon (after doing the convert, to do
things in more or less the right order...)

> I hope that my English is understandable.

Sure it is ;-)

Regards, Barend

Geometry list run by mateusz at