Geometry Boost List: [ggl] Problems with the difference between two polygons

Subject: [ggl] Problems with the difference between two polygons
From: Enrico Leoni (lions.forever)
Date: 2011-07-31 15:22:25

Hi Barend,

On Fri, Jul 29, 2011 at 1:16 PM, Barend Gehrels <barend_at_[hidden]> wrote:
> Thanks for including Boost.Geometry in your benchmark. I was able to
> reproduce your benchmark, thanks for the sources, and played with it.

Thanks to you, I'm glad to know it was of help.

> 1)
> So my question is: can you check it with this define? It is in the Trunk now
> (so it is just added/committed, and skips two lines of code). I'm curious to
> the new results (I didn't run all tests for all libraries on my machine, and
> I didn't compare or create graphs). I will check on our
> self-intersection-check later, that will take some time.

I will check it in the next few days (the pc used to run the tests is
located at my workplace).

> 2) It would be great if you could add Microsoft's SqlGeometry to your
> benchmark.

SqlGeometry has been captured by my radar some weeks ago (reading some
of your blogs indeed) and I have already added it but before posting
something I will have to check a couple of things.

> 3) It would also be great if you could add GEOS to your benchmark. This
> would require adding another C++ wrapper.

I just tried to run some examples compiled with the /clr flag and all
seems ok so it is very likely that GEOS will be added too.

> 4)
> You might consider to add your benchmark
> sources to that repository. It is nice to have a benchmark both in pure C++,
> and in .NET, available. You also might see how I did e.g. GEOS.

With SqlGeometry (and probably GEOS ... while CGAL has some problems
when compiled with the /clr flag) I think I will close with this
benchmark ... also I think it should be at least refactored / refined
(to keep the myth of the VB programmer I did a lot of copy and paste),
and the wrappers should be created using P/Invoke (maybe with the help
of SWIG) for portability (Mono under Unix). This is more work than I'm
willing to do. However, if someone is interested, my code is available
and can be taken and used without problems.

Regards, Enrico