[OSM-dev] Mapnik vs Osmarender (was: A proposal for linking multiple ways...)

[Jon Burgess]

On Thu, 2007-05-17 at 01:42 +0200, Frederik Ramm wrote:
> Hi,
> 
>     just going off on this tangent:
> 
> > I'm sure the coastline rendering would be more consistent if we
> > developed something which was used by both mapnik and osmarender.
> 
> As I have said before, I personally view osmarender/tiles at home as a more 
> flexible, more playful, almost "rapid prototyping" thingie where we can 
> develop ideas and quickly try them out, and Mapnik on the other hand as 
> a more heavyweight, more formal, more "correct" solution where we can 
> introduce concepts that have proven useful.
> 
> In the long run, we will have one or more Mapnik-like central renderers, 
> I am pretty sure about that; but meanwhile, holding back development in 
> tiles at home until solutions common to both worlds can be found would kill 
> off one of the main advantages of tiles at home.
> 
> Bye
> Frederik
> 

It is interesting how you see the two renderers. 

The funny thing is, in my experience, I see the opposite. I think it is
quicker to setup the mapnik tile rendering and faster to re-render a
large set of tiles in a new style than osmarender/t at h (based on some
very limited experience with using the XSLT style transforms some months
ago).

The main mapnik/tile setup may be more centralised and appear less open,
but to me it seems faster to prototype stuff on my home machine with a
new mapnik osm.xml file and a local postgres DB based off planet.osm.

Using something like tilecache I can have my own local slippy map which
I can re-render in real time as I zoom around the globe.

	Jon