On 11/11/12 09:31, David Bannon wrote: > Andrew, thanks for the very carefully considered response. > > I agree with just about all the points you make but suggest your > conclusion may not quite address the driver for this proposal. This is > about getting knowledge into the hands of end users. Safety is a key > factor, I am sure you heard someone died on an outback track only two or > three days ago.
I can defiantly relate to the driving force here. I myself have looked at the osm mapnik map and planned a route based on it, only to find myself reaching unexpected dirt (or worse sand) 20km into the road. I wasn't prepared for the road becoming unpaved, but the tragedy was that the road was actually accurately mapped out with surface tags, I just never downloaded the data before the trip to find which sections were paved and which were unpaved. This is a complete failure of the cartography and if it represented unpaved vs paved as dotted casing then I would have been prepared and expecting the surface change along the road. > > What I am trying to do is get a reasonably easy to understand and use > model in place so mappers put data into the database in a consistent way > and, critically, the rendering people and the routing people use it. At > present, we have a large number of tags that relate to this space but > the only one the renderers use is tracktype. They ignore > surface=unpaved, 4wd_Only, smoothness= and so on. > > Your suggestion, that we focus on specific characteristics of the road > and describe them is a good thing and one I'd support if you were to > start working on it as a proposal. But it does not apply here. > > While having breakfast I could easily think of ten items I'd like to add > to your list of road characteristics, I'm quite sure that if we sat a > few more 4wders around the table we could hit twenty in no time. If we > convert them all to tags we'd have two real and pressing problems - > > 1. Mappers could not cope with that many tags to address in a consistent > way, even if they did, think of the extra data that would be, in many > cases unnecessary. Subsequent validating of such a fine grain data would > be a daunting task. > > 2. Importantly - it would be impossible to get the rendering people to > even consider displaying such a range of data. And if the mainstream > rendering engines don't show it, the routing people ignore it. Truth is, > when an end user uses OSM to decide to go a particular way, he/she does > not fire up JOSM and examine the tags on each way, they look at a map, > or, worryingly, rely on a routing tool. http://wiki.openstreetmap.org/wiki/Tagging_for_the_renderer Although I can synthesis with this, at the end of the day, osm projects build the database, not a cartographic map. You are free to make your own rendering style which represents surface tags, similarly you are free to define your own routing rules which determine how a routing engine chooses a route. > > I believe we need something simpler, something that is an extension of a > tag that is already rendered, for example on the OSM website. Yes, > tracktype does focus on compactness at present but thats because in the > range of roads it currently addresses, thats the issue. I propose we > extent that range of roads. I think your extension proposal make is more complicated as it is unclear what the scale represents since it isn't a linear scale for one attribute. Admittedly the current tracktype tag also suffers from this problem, but the wiki page does suggest the two attributes represented separately as surface= and smoothness=. It does seem to me that accurately taging surface (gives you an indication if you need a 4wd for traction; --I'm not a 4wd user, so I may be completely wrong about this), and smoothness (if you need a 4wd for clearance) would sufficiently meet your requirements. If only it was that easy... We have, http://wiki.openstreetmap.org/wiki/Key:surface http://wiki.openstreetmap.org/wiki/Key:tracktype http://wiki.openstreetmap.org/wiki/Key:smoothness http://wiki.openstreetmap.org/wiki/Proposed_features/surface_unification http://wiki.openstreetmap.org/wiki/Proposed_features/usability http://wiki.openstreetmap.org/wiki/Proposed_features/mtb:scale http://wiki.openstreetmap.org/wiki/Key:trail_visibility http://wiki.openstreetmap.org/wiki/Key:Sac_scale http://wiki.openstreetmap.org/wiki/Tag:4wd_only%3Dyes > > WRT the mention of "bad data that needs to be fixed", you must note that > this proposal does not make that issue any worse, indeed, by focusing on > one linear tag, tracktype, it might actually help. The problem is not > strictly linear but can be projected onto a linear tracktype scale at > the granularity proposed given the sort of knowledge any reasonably > experienced 4wder has. Importantly, it's better than nothing and that, > I am afraid, is the likely alternative. Although this issue does affect Australia due to the nature of the outback, it is a global issue. I think it would be best to take your thoughts to the global tagging list at let the discussion happen there.
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Talk-au mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk-au
