[Gdal-dev] Re: ogr geometry intersection

[Jack Riley]

Hello Todd:  Thanks for the information re: the disparity between the
geometry constructions and tests in ogr (geos).  In terms of visible-extents
clipping (draw / do-not-draw decisions), I suppose a small expansion of the
clipping rectangle will bridge the gap in precision.  For more stringent
topology requirements (e.g., strict point-in-polygon tests that guarantee
mutual exclusivity between adjacent polygons) the disparity should not be
ignored.

Frank & Mateusz:  Thanks for looking into the ogr performance issue.

Jack


On 3/23/07, Mateusz Loskot <mateusz at loskot.net> wrote:
>
> Frank Warmerdam wrote:
> > Jack Riley wrote:
> >> It appears that the FWTools v1.0.9-and-earlier TopologyException is
> >> limited to situations where one of the two polynomials being tested
> >> for Intersection() is completely within the other.  In other words,
> >> the Intersection() test in "old" versions of FWTools/GDAL works--and
> >> works "fast"--if the two polynomials actually intersect.  So, my
> >> workaround is to use FWTools v1.0.9 (or earlier) and condition any
> >> call to Intersection() with Disjoint() or Contains() test(s).  The
> >> question remaining is:  can FWTools v1.1.0+/GDAL revert back to the
> >> use of the "old" and "fast" Intersection() code, adding in a test of
> >> Disjoint()/Contains() to circumvent the TopologyException?
> >
> > Jack,
> >
> > I've asked Mateusz who is a GEOS guru too to look into this issue.
>
> Jack,
>
> I'm going to try to reproduce your results and tests today and tomorrow.
> I will report about results.
>
> Cheers
> --
> Mateusz Loskot
> http://mateusz.loskot.net
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/gdal-dev/attachments/20070326/1edd455e/attachment.html