[Routing] SVG instead of tiles was: Routing web-services

[Robert (Jamie) Munro]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

OJW wrote:
> On Tuesday 20 November 2007 16:57:13 Robert (Jamie) Munro wrote:
>> In all off-line cases, we should be downloading vectors and drawing the
>> map on the client. They are much smaller, and they can be scaled,
>> rotated and 3d projected, while keeping the text facing in a sensible
>> direction. They can have any elements hidden or displayed, then can be
>> directly routed over without needing another set of data.
> 
> well show us some vector map viewers that outperform image ones then... all 
> that scaling, rotating, and 3d projection (on every node in view, and lots of 
> nodes not in view) all takes time.

How about Mapnik? Or every device made by Garmin, TomTom, or included in
a car. Or look at computer games.

> it's not even the same data as you want for routing.  The routefinder doesn't 
> care what *name* a street has, only how fast you can get down it with a herd 
> of camels.

It's highly overlapping data.

> I guess the tiles at home project should be able to tell us a bit about the ratio 
> of image size to OSM data size to SVG size for each tile.  And about the CPU 
> power required  (last I heard, it was 2GB minimum RAM and quad-core CPU for 
> rendering an SVG of central London)

That's because it's doing several thousand zoom 18 tiles in one go, and
because XSLT / SVG

> I did look at using the vector data for display in pyroute.  And I got results 
> looking similar to all the other vector map displays -- i.e. that the maps 
> are very slow and very ugly.

Have you tried using the mapnik python bindings?

Robert (Jamie) Munro
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHRACaz+aYVHdncI0RAmJtAJ0Qtr/RtZ5nUXkFHemb0rznv35ZbACg0MEg
1VAQ3byIRj1GQnqDuKyuVyk=
=P6BT
-----END PGP SIGNATURE-----