One thing I don't understand about QuadTiles is the use of base 4 data. For example, if A=NW, B=NE, C=SW, D=SE (just like at http://wiki.openstreetmap.org/index.php/QuadTiles), then "ACB" represents a large chunk of the western USA.
This makes sense. However, if convert 00=A 01=C 10=B 11=D, we get: 000110 (base 2) This doesn't make sense, because that's really the same number as: 00000000000[...]000110 (base 2) In other words, we're using "00" to mean "NW corner", but forgetting that "01" and "0001" are equal as numbers (they're different as strings, but equal as numbers). Should we store an extra bit to record the granularity/size of the QuadTile? My thought: use base 5, with 1=NW corner, 2=NE corner, 3=SW corner, 4=SE corner. So, "ACB" would become 132 (base 5) [42 in decimal]. Of course, "1320", "13200", etc (in base 5) would all be alternate reprsentations of this tile (we now have multiple representations of a tile, but that's better than ambigious representation). To get tile inclusion we'd divide by 5 and throw out the remainder. Vaguely related thought: do QuadTiles have to be "square"? How about using base 9 with: 1=NW corner, 2=north half, 3=NE corner, 4=west half, 5=east half, 6=SW corner, 7=south half, 8=SE corner (and 0 used as above). This would lead to tighter (rectangular-shaped) bounding boxes, and we'd still have the "divide by 9 to get inclusion" property. Note: We could get rid of all ambiguity by writing QuadTiles as decimals (eg, ".132" and ".132000000" are the same number), which might be worth considering (for storage, we could multiply the decimals by 2**32 or 2**64 or something) -- We're just a Bunch Of Regular Guys, a collective group that's trying to understand and assimilate technology. We feel that resistance to new ideas and technology is unwise and ultimately futile. _______________________________________________ newbies mailing list [email protected] http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/newbies
