El Miércoles, 22 de Abril de 2009, Frederik Ramm escribió: > This does not rule out the application of a convex hull but [...] > It is going to be expensive, and it is not going to avoid false positives > altogether, just reduce them...
My point is that we could reach a compromise between no false positives and computing time. Besides, computing a bbox is O(n) (n being the number of nodes affected by the changeset) and a convex hull is O(n log(n)), which isn't that bad. Geez, I do need to improve my computational geometry skills ASAP. There is just so much cool stuff to be done. > I think an external service that does nothing but compare > "subscriptions" to changeset bboxes and the elements therein could be > the solution; it can take all the time it needs, and maybe even apply > some automatic segmentation to large changesets. This whole 0.6 changeset stuff just turned into "ITO OSM mapper is dead, long live ITO OSM mapper!" :-P -- ---------------------------------- Iván Sánchez Ortega <[email protected]> Palabras Textuales #603: "Estos valores de CMYK no pueden estar bien por que si los sumas todos no da 100%" (Agencia, Departamento de Producción a arte finalista)
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ talk mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk
