Dane Springmeyer

I'm not sure how to help ruby work directly with `long long`. But we can fake it with `int` which I've done in 39e76e0 and that gets the test passing....

Add `mapnik.register_datasource(path.join(mapnik.settings.paths.input_plugins,'postgis.input'));` or just `mapnik.register_default_input_plugins()` at the top of the js code. The reason is that you are running latest node-mapnik and the examples have not been updated to account...

btw, you might be interested in https://github.com/springmeyer/coalition. It is the first iteration of what later turned into the cartodb tile api.

Oh, so I guess I thought wrong. If you don't have the function `mapnik.register_datasource` then your node-mapnik is not bleeding edge. So then you must actually be missing the Mapnik...

Feeding in a Mapnik XML is the right way to generate vector tiles. You are correct that the `style` information in this case is ignored - however there is no...

I'm seeing a crash running the tests: #46

@rundel - yes, this definitely needs thought. Hopefully @herm is amenable to keeping the deprecated text_symbolizer functions (that should also work for the common use case) for the time being...

this solves the two issues I encountered where the output json was not valid, but surely we should find a better overall approach to escaping: ``` diff diff --git a/include/gen_layer.hpp...

@rundel - any thoughts on this one? This is the only major blocker to converted the massive "osm.xml" at http://svn.openstreetmap.org/applications/rendering/mapnik/

Interesting. Not seen this on osx. I'm now also running boost 1.48 so I will try to test again. @rfw and @rcoup - you guys tried to run carto-parser recently?