https://openstreetmap.org/copyright | https://openstreetmap.org |
Copyright OpenStreetMap and contributors, under an open license |
https://openstreetmap.org/copyright | https://openstreetmap.org |
Copyright OpenStreetMap and contributors, under an open license |
I think the author was new and didn't know how to define highway areas. OSM has had differing propsals of how to do them correctly. This area is very difficult to route padestians from the very narrow footpaths leading away from the roads and the very wide roads so area highways are the only way to avoid needing to bodge the interfaceing between them in an albut standard logical way. scale effects are not helpful. Some work on defining the areas was started but offset variations and a lack of time and proper survey measurements have delayed this section. It needs a set of measured refences to under a meter in hoizontal accuracy to be done right.
In cases of footpaths (sidewalks) that are effectively part of the road, sidewalk=left/right/both is surely the better way to map it?
I'd also argue that road-as-area mapping isn't something that "works" in OSM yet - we need a different paradigm akin to "river" and "riverbank".
This area is on a long todo list that is steadly working through the whole estate fixing the old mapping style battle sites and using the now resolved mapping standards for areaisations of highways.
The problem was one of scale for makeing parking and predestrian to entrance mapping as the sidewalks morph into sidewalks to wide roads and then off into the equivilance of residential roads that the AtoZ has begun to name I think as resdential footpaths (or was it footways?). The new tagging scheme back then experimental can deal with all this.
I'm not sure what the "now resolved mapping standards for areaisations of highways." are. There are at least three "competing standards" - highway with are=yes (which you've used on http://www.openstreetmap.org/way/244879222 ) , area:highway (see http://wiki.openstreetmap.org/wiki/Proposed_features/area:highway ) and landuse=highway (see http://wiki.openstreetmap.org/wiki/Proposed_features/landuse%3Dhighway , which is a solution to a slightly different problem). You've gone with area=yes on what is essentially a linear highway=residential . Unfortunately, this breaks routing (how can a router tell a "genuine" highway area from a fake one like this) and leads to rendering anomalies due to the duplication you've introduced (see http://c.tile.openstreetmap.org/19/260176/174640.png ). See also http://wiki.openstreetmap.org/wiki/Key:area#Highway_areas for why this isn't a good idea. Personally I'd have thought that mapping as "area:highway" (see http://wiki.openstreetmap.org/wiki/Proposed_features/area:highway ) would have been preferable, if only because it wouldn't break anything else.
However, mapping of highway areas is very much in its infancy - I'd suggest it's worth discussing on talk-gb (rather than just in changesets and osm notes) what the best way forward is.
Cheers,
Andy
well this system isn't geared up to the talk-gb automaticly but try find the links in a while. but as reader of this comment spot might be intreasted I'll keep it very brief.
Basicly I've been trying to figure out amongst the diffeent systems for more than a year or maybe two now and this estate is a place I tried to figure them out on. Its not just expirimenting but a genuine atempt at what seemed the current way to do things at the time I did them.
Meldrum Close is a case were things are differnent to how I do them now and after mentor wadeing in with a another choice on areaisation than what I hade put and things getting meny names it really is a mess at the moment landuse=highway works like landuse=railway & landuse=Transport which are respective areas covering a whole space dedicated to those subjects like landuse=education etc. it can show the effective boundries of highways and other transport systems when combined with operator or owner tags can define what is adopted and what is rented for public use etc (see road widening schemes espcialy near some shops and other public aminities that put the footway over there frontgardens for comercial benifit).
Mentor misused my landuse tagging to define an area as if it was the whole carrageway and didn't adjust Villiers Walk wich is why some of the rendering fails so badly look at the other end of the walk thats been properly fixed to see see how the residential_footway ends in the sidewalk_footway before using footway=sidewalk tag to join and cross the sidewalk before changeing into a crossing_footway across the carrageway before changing back to a sidewalk_footway routeing line to complete the journey to the sidewalk_footway routing line on the sidewalk on the opersite side of the road this is logicaly correct to were the line is and helps the router algorithium know this is a logical to place to cross and what places to stop people to use the green cross code it also often gives a cleaner render. Wilmot Walk sufers abit as Osmrender feels it should be wider so it isn't rendering the area but only the routingline to get a enough for the minimum width and the caps style give the bulgey ends with the footway of Stapleton close but not with the similary line only Enbourne Street yet to be areaised.
Meldrum close also suffered a misalignment problem with the pictures in bing and the original question poised in the notes was which was correct could someone measure it for us please.
As for standards other than landuse. I was originaly in the area:highway camp but both mentor, osmrender and others seemed to want to use area=yes form and there was a logic as other OSM features work this way and they would change my area:highways=* and for a bit in the early confusion highway:area=yes and highway=area even to the area=yes form wether I liked it or not; so it fitted there systems.
The problem with using the standard area=yes format is in other parts of OSM things aren't defined twice by highways gett the simplifed routing lines which is why area=yes dosen't work well for a lot of simple searchers.
The comprimise was to do both systems similtainiously so that people would fight over which system to change it to next and offer but camps of imbeded user the opertunity for algorithiums to easily spot a area form instead of a routing form {basicly if I did it recently and it has a tag like k=area:highway then treat it as not the routing line regardless of anything else - this is easy in mapcss and other boolean patern matchers}.
Finally on behalf of both myself, bing image setters and mentor the reslting Meldrum close is a bit of a "dogs dinner" of a mess for which, at least, I for my part appoligise for I hope that when not on other spots on the map and not writing discourses in comments I may reach that bit of the estate. Note I dontthik I finished Skipoons close yet as lot of footway routing lines need improving and the walk may need to move to a new discription probly after the AtoZ nameing of this type like highway=footway footway=residential
This thread is quite old and highway area mapping has moved to area:highway=* since the initial mapping. We also now have OSM UK cadastral parcel layer. So I have updated the mapping here to reflect both.