I was thinking more the really rough TIGER roads around the oilfield - most rural TIGER highway=residentials are an approximation at best, but this bunch is really bad and actively misleading. If all the oilfield roads were deleted, and a few access roads traced afresh, the map would be a lot better. But you're right, the landuse doesn't help!

I'd certainly prefer cycleway=separate. But I can take it to the tagging@ list if you think that'd help.

SK53 has pointed out that sidewalk=separate is frequently used (https://taginfo.openstreetmap.org/tags/sidewalk=separate, https://github.com/graphhopper/graphhopper/issues/712, etc.) to indicate "road has a sidewalk, mapped as separate way".

By analogy, cycleway=separate would be good tagging to use in this case.

One feature, one OSM object is a well-established principle: osm.wiki/One_feature,_one_OSM_element

In this case, if you have two representations of the same object (a cycleway), one with a less accurate geometry than the other (i.e. on the road centreline rather than on the actual cycleway), then the user has a 50% chance that the polyline in their turn-by-turn directions will send them up the inaccurate geometry (a busy A road) rather than the accurate one. Which is clearly not a good thing!

By and large you should not expect OSM data (or, indeed, any detailed geodatabase) to be directly consumable for small-scale mapping, but should expect to have to do some generalisation. OSM tends towards completeness and that means the data gets more detailed and less directly consumable for small-scale maps as time goes on.

If you were to create some proprietary tags to represent this scenario in a format more readily consumable by your rendering rules (a la network=xcn in Oxford) then I wouldn't object. I wouldn't even go to the barricades against an _additional_ tag accompanying cycleway=track, to express "the track is mapped separately", though it would be suboptimal and contrary to mapping practice the world over. But having the same feature represented by two different geometries, without any indication that they are the same feature, is not sustainable.

Hi Mike,
That's great - thank you. Went for a walk today and it is still (happily) very much in evidence on the ground, albeit a bit muddy!
All the best
Richard

Hi Mike,

Interested to note your retagging of the D'Arcy Dalton Way with a comment that it's "no longer classified as a long-distance path". It still appears to be signposted round here and the guidebook has just (December) been republished. Can you shed any light?

All the best
Richard

Just because a document doesn't have a copyright notice on it doesn't mean it's uncopyrighted, I'm afraid. Quite the opposite- copyright subsists unless expressly disclaimed.

Second, a document being "accessible" has no bearing on its copyright status.

Third, as SomeoneElse explains, the consensus in the UK is that we do not add C/D/E/F/UX/etc road numbers to the ref tag, but rather to official_ref or somesuch (if they can be legally sourced). The problem with your (rather sweeping) "bad product" assertion is that OSM is an international project and it is not encouraged to unnecessarily add regional exceptions unless there's a compelling reason. ref= means the same thing the world over: let's not add a needless exception for one country.

I can't fault your enthusiasm, but these aren't tracks ("roads for mostly agricultural use, forest tracks etc.; usually unpaved (unsealed) but may apply to paved tracks as well, that are suitable for two-track vehicles, such as tractors or jeeps"), and they're too transitory to map in OSM. Sorry.

Looks great! Should this be network=rcn instead? I think (but might be wrong) it's only the USBRS routes that get network=ncn - osm.wiki/WikiProject_U.S._Bicycle_Route_System

Geograph photos:
http://www.geograph.org.uk/photo/408432
http://www.geograph.org.uk/photo/1036267
http://www.geograph.org.uk/photo/408921
http://www.geograph.org.uk/photo/408923
http://www.geograph.org.uk/photo/4019525
http://www.geograph.org.uk/photo/408927
http://www.geograph.org.uk/photo/407050

Hm, yes. I downgraded it from route=bicycle to route=mtb, so I'm not adding any incorrectness that wasn't there already :), but I guess it probably needs two relations (one mtb route, one walking route).

Seriously, there is no bridge over the Severn at Saxon Lode. There was once, but it's been demolished. I've been past a dozen times on our narrowboat and I would definitely know if there were a bridge there. Perhaps be a bit more careful adding features from historic OS maps? :)

Thanks!

(By "this", I mean relation/6196005 of course - I should have made that clear!)

I think there's an issue with having this as an otherwise untagged multipolygon. "Old-style multipolygons" (tags on the ways, not on the relation itself) are still common in OSM. Given the lack of substantive tags on the relation, this could potentially be interpreted as an old-style multipolygon, leading to a grand kerfuffle with a road following the outline of the park - indeed osm2pgsql will do that under certain circumstances.

Since this relation appears to exist for archiving purposes only, could I suggest either deleting it, or tagging it as something not a multipolygon? (Perhaps type=collection)

Hi!

Please don't add names like "NCN 7" to paths. In OpenStreetMap we record that a path is part of NCN route 7 by making it part of a "route relation". This shows up on the OpenCycleMap layer and other bike-specific cartography.

Adding it to the name as well is duplicate information and makes it difficult for cycle routers and renderers to show the correct name.

Thanks!

Richard
(cycle.travel and Sustrans ranger)

It shouldn't really be either, tbh! Nowhere else in the world is a mountain bike route tagged as route=bicycle - that's what route=mtb is for.

I've pretty much given up arguing this one but I wish at least people could agree on a static (albeit wrong) network/ref combination - I've now got to spend another 24 hours regenerating my US routing graph because the tagging change has broken my previous override. :(

(Sorry, not directed at you particularly, Steve - I realise there are some other strong-willed people editing this relation...)

@Lutz: it would be kind if you were to write in English when changing data in England mapped by English mappers. Pointing to a German-language wiki page is not really very helpful.

Hi. Mousehole is a lovely place. :) I wouldn't tag it as 'waterway=dock' though - that's for a place where the water level can be controlled (e.g. by a lock), and Mousehole is entirely tidal. Look at the Bing imagery and you'll see it's low tide and the boats are all sitting on the bottom!

Ah yes! Fixed. Thanks for spotting it.