> > Other ideas about performance improvements are welcome. For instance, I > have had some ideas about only noding geometry in the region of actual > overlap in the case of an intersection computation. That probably has > implications for robustness, however. Needs some research... >
I tried to implement this speed up in PostGIS via box clipping. Current GEOS recangle clipping is unfortunately not robust, but helps quite a lot when not crashing. I tried redoing clipping by doing just O(N) clamping to the box. Result retains many required properties (ie, raycasting will give correct inside/outside result for it, area is okay) but has holes touching edges and edges touching themselves. GEOSIntersection can't handle that. Is there a simple way to recover the boundary for such geometry to feed further to GEOSIntersection? Buffer(,0) does not help in many cases, ST_MakeValid too. Can we make GEOSIntersection handle such invalid input, or is it completely incompatible with monotone chain conception? -- Darafei Praliaskouski Support me: http://patreon.com/komzpa
_______________________________________________ geos-devel mailing list geos-devel@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/geos-devel
