I'd like to pick the brains of those of you who are better at maths,
geometry etc. than I am (probably all of you).
At present, P2 sends a /map request to the server every time it needs
data for the current viewport. It does cache data (subject to not
boggling Flash Player by keeping too much around), so if you return to a
previously loaded area, it won't issue a /map call again.
But if the user is in a densely mapped area, and pans (say) a couple of
hundred metres northwest into an area not currently loaded, P2 will
issue a /map call for the new viewport.
It would be much better if it issued two smaller /map calls: one for the
northern edge, one for the western edge. That way, the 90% of data
already in memory wouldn't be rerequested.
I'd love to code this but figuring out the maths is beyond me. So can
anyone supply pseudo-code[1] that:
- given a - an array of already loaded bboxes (each one being left,
right, top, bottom)
- given b - the bbox we now want to load
- given n - a maximum number of /map calls to make (e.g. 3)
- returns c - an array of no more than n bboxes, covering the areas
within b which are not in a?
cheers
Richard
[1] real code even better ;)
_______________________________________________
Potlatch-dev mailing list
Potlatch-dev@openstreetmap.org
http://lists.openstreetmap.org/listinfo/potlatch-dev
