Hi Marko, I had the same idea while I was optimizing splitter, but I wanted to achive better throughput. I don't know if it also would allow better quality. Regaring the performance of the LocationHook I see no advantage doing this. I think WanMil introduced the precompiled bounds just because the normal input doesn't contain enough information (unclosed boundaries). Ciao, Gerd Marko Mäkelä wrote
On Sun, Jan 08, 2012 at 10:36:13PM +0100, WanMil wrote:
I broadly remember that splitter uses a 2^n raster (?). Maybe it's beneficial to change the boundary raster to the splitter raster?
Would it make any sense to combine splitter and mkgmap to a single process? I guess that splitter would still have to write output to disk in many cases, but in others it could directly pipe the tiles to the mkgmap code. Would it help if multipolygons and location info were processed directly in splitter?
Marko _______________________________________________ mkgmap-dev mailing list mkgmap-dev@.org http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-- View this message in context: http://gis.638310.n2.nabble.com/Bug-in-LocationHook-tp7157897p7167257.html Sent from the Mkgmap Development mailing list archive at Nabble.com.