Felix Hartmann wrote:

I do agree though that OSM's tagging for road surfaces is a bit of a mess, but it needs an OSM-level cleanup if that's a problem, not at mkgmap-level.

AFAIK there are "surface=???" "smoothness=???" "mtb:scale=???" "sac_scale=???" "rtc_rate=???" tags in OSM, all of which (sometimes in combinations) ought to be enough to give mkgmap the clues needed to set the routeability of a given way. Plus "access=???" and "<vehicle>=no" of course.

Not just that, but those tags already exist. We should be using them.
Steve

You don't seem to understand.

No - I understand fine, thanks.
I just disagree with the idea that this proposed new tag should be mkgmap-specific.

You can use them and the default style should use some of them. However there is no clear borderline of what is paved and what is unpaved, therefore it is best to use a new key.

Well, a new key is one option, getting the entire OSM community to tidy up a bit on the use of existing keys is another option.

IMHO, if use of the existing keys isn't feasable, then any new key should be more like "autorouter=avoid" or something. This is not mkgmap-specific, it would apply to all the other projects looking to use OSM data for in-car GPS navigators, and the browser-driven computer navigators too.

But before adding a new key, we should be able to "prove" that no combination of the existing keys does it for us.

Steve