[OSM-dev] Improvement of tile rendering

Udo Giacomozzi udo.osm at nova-sys.net
Thu Mar 19 23:49:43 GMT 2009


Hello Dane,

Wednesday, March 18, 2009, 12:42:14 AM, you wrote:
DS> Cool, sounds like it works fine though and many people use PHP so you
DS> might considering adding your script to
DS> http://code.google.com/p/mapnik-utils/ 
DS>   since others might find it useful.

It's really just a simple PHP script that invokes the Python script:
http://paste.debian.net/31026/

single_tile.py is a modified generate_tile.py that renders only a
single tile based on three environment variables:
http://paste.debian.net/31027/

As said, this was the first time ever I came in touch with Python, so
the patch is really simple. ;-)

Anyway. This method keeps the server busy with invoking Interpreters
and parsing scripts instead of really do something useful, that's why
I would not suggest it for any production environment.

Instead I'm thinking about writing some multithreaded FreePascal or
C++ FastCGI server that does it much more efficiently (by using
libmapnick internally) but I still need to have a look at mod_tile as
it might already provide a highly performant solution.


DS> Stefan pasted his python/cherokee solution, so I'll offer mine too. I
DS> like to deploy TileLite using mod_wsgi, and for anyone interested the
DS> code can be found here:
DS> http://bitbucket.org/springmeyer/tilelite/

The features list sounds promising! How does it compare to mod_tile ?


DS> Given mapnik trunk (or the upcoming 0.6.0) a 'map.buffer_size' can now
DS> be used to help avoid cut labels instead of this view based
DS> approach. (http://trac.mapnik.org/browser/trunk/CHANGELOG#L64 
DS> )

Where is buffer_size documented? How does it help avoiding cut labels?


Thanks for your hints,
Udo






More information about the dev mailing list