[Tagging] Comparison of tag support: Mapnik, Osmarender, Potlatch, JOSM, Kosmos, Map Features (wiki)

Steve Bennett stevagewp at gmail.com
Mon Dec 14 10:59:15 GMT 2009

On Mon, Dec 14, 2009 at 9:02 PM, Pieren <pieren3 at gmail.com> wrote:
> What is "core" and what is not ? What is in the wiki Map Features ?

A subset of those, probably. There is some really esoteric stuff in there.

> But anyone can add his tag, approved by 10 or 12 voters in the Map
> Features...
> You cannot fix priorities for other people creating the data or other
> applications using the data. What is the "core" for OpenCycleMap is
> not the same as the "core" for OpenWaterwayMap, OpenCrimeMap,
> OpenHikingMap, OpenShopMap, OpenTourismMap, OpenSchoolMap, etc...

As I think I said, I agree that different renderers *with different
purposes* (which is what you're referring to there) should implement
different tags. But these cases are not good:
- different interpretations of the same tags
- arbitrarily different sets of supported tags
- renderers that support deprecated tags better than they support
recommended tags
- renderers that do things differently enough that people end up
targeting a particular renderer.

Btw, I do think that OpenHikingMap should interpret highway=cycleway
exactly the same way as OpenCycleMap, OpenSwimMap,
OpenMarijuanaMap...etc. They will each render it differently, but it
should *not* be the case that "cycleway" means a paved surface in
OpenSwimMap but it means a mountain biking trail in OpenHikingMap.

Does that make sense? It's pretty important.

> You are really expecting too much from the "rendering" side of the
> project. What we need is a clear definition of the tags and this is
> done by searching a consensus on the wiki and on this list and we see
> if we succeed or fail with the statistics.

Absolutely. I'm only constantly referring to the renderers because
they're the physical manifestation of that definition. There may be
other uses of OSM data, but the renderers matter the most.

> If we don't have a common
> definition of the tags, we end up with different rendering rules and
> even worst, with different tagging presets in the editors.

Yep. Yep. Yep.

> So, what we have to do first, if we have to fix priorities, is to
> improve the definitions of the tags, avoid duplicates or changes in
> the meaning of the tags one year or two after their creation (e.g. the
> other thread about cycleway on this ML).

Yes. Part of the process fixing a tag may involve mass updates to
existing data, or, more painfully, deprecation and a slow, manual
check and update.


More information about the Tagging mailing list