I have made a version of dime that compiles with gcc on linux and has openstreetmap output.
It uses a hacked version of proj, which i want to integrate closely using c++ templates to reduce overhead. It is my opinion that we can add in all these templates to produce a inline template function that does the conversion, this will be able to be optimized very well. here is my diary entry about it: http://www.openstreetmap.org/user/h4ck3rm1k3/diary/9137 example data (OUT_1.osm.bz2) files from brasil : http://ia341329.us.archive.org/1/items/VitoriaOpenstreetmapData/ Code is on github: dime : http://github.com/h4ck3rm1k3/TwoNickels proj.4 : http://github.com/h4ck3rm1k3/ProjC--/ I want to use my new template classes with all these tools to better integrate them. I can imagine that there is one point template that maps all points in all the tools to one register of memory. Meaning that we can replace the types in the older system to synthetic template types that present a facade to the existing interface. Also you can read my diary entries on the EPANatReg, the epa which is a high speed sax2 xerces-c template processor. I have written lots of little command line processors that are very fast. You can make template callbacks access world data in a high speed custom way. a little bit slower that wc, the wordcount. https://code.launchpad.net/~jamesmikedupont/+junk/EPANatReg Happy new year. mike _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/listinfo/dev
