On Tue, Jul 8, 2008 at 11:49 AM, Alex Wilson <[EMAIL PROTECTED]> wrote: >appropriate interface on them, you can then use all the algorithms whilst >retaining full control >over the space requirements. It's simply not true that >there is an unavoidable large space >overhead.
So you must 1. Pack the data in yourself esp. if you want to avoid cache misses / paging. 2. hook a callback for testing to see if a segment may be traversed and at what speed. 3. How do you prevent a bad routing query on a large map from eating all the memory on your WinCE device causing a crash ? 4. If the query is impossible (split network), can you modify the algorithm to at least return some useful route to the user ? > Finally: why would you want a library that combined routing and rendering? > They're two separate tasks: what advantage is there to combining them? Well you want to use the same data for rendering and routing on an embedded device with little memory. For rendering you want to be able to find all nodes in a given region (rectangle) and for routing you want to be able to take one node of a segment and find all the other segments at that node. So the two operations are independent except for performing different queries on the same data structures. _______________________________________________ Routing mailing list Routing@openstreetmap.org http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/routing
