Rest of mine done as 34979326

public_transport=

These are the end points of the actual course of the bus along the route. This one corrected in changeset 34979068. I'll have to look for the others.

There are actually some more paths. There are exits onto both roads at the SW corner,s so a fork there. Also the path at the southern tip continues straight on to join the circular part about halfway round, as well as forking to cross towards Arc House.

There is also a path around the NE face of Arc House, which then hits a T junction, with a dead end to the NE and leading into the stage and terrace areas to the SW.

I took a GPS trace osm.org/user/hadw/traces/2019500 b but it is quite noisy. This starts at the S, joins the main circle and goes anti-clockwise. Goes up to the bollards and cycle stands, then back on the main circle up to the SW corner of Arc House. Then follows as close as possible to Arc House going clockwise to the end of the stage are. Goes back along the outer edge of the stage area then does a loop of the centrelines of the stage and the pit below it and back onto the main path. Next it visits both SW exits, and again continues with the main path. until the intersection, where it turns left, makes a detour to the centre of the basket ball practice area, and finally rejoins the main path. It completes the circuit of the main path, backtracks the original entry path, until it hits the diagonal path, then heads back to the basketball area, where it tries to trace the perimeter of the court.

The sign says operated by Harrow College, and also says: events, dance, music, cafe, open daily, but it wasn't open around 14:30 today.

Arc House is rectangular, and significantly elongated and I wouldn't classify it as "commercial", although refining the full nature probably awaits its opening. My provisional guess is that it is an open air theatre.

osm.org/way/359274173 is tagged as a path, but use on foot is forbidden. It is also tagged with mtb: tags, but use on pedal cycles is also forbidden.

More generally, please always give an explanation of the purpose and scope of your edit, and the source of all the data used, in the changeset comments and tags.

addr:housenumber should contain only a number and that should be the number on street. Check the Thai community for how to number offices in a building but addr:unit is common.

Unless the Thai mapping community has strange rules for addresses, the address should be in the language that would be used for addressing a letter, presumably Thai.

This changeset has caused rendering issues in some renders, but renders correctly in Mapnik standard layer, the GeoFabrik layers uses by the multipolygon checker tool and in the editors JOSM, Potlatch 2 and Id.

It does fix a rendering issue with the island not rendering, because the water took precedence, and I believe it more correctly describes the true topology.

I haven't tried layer 1 on the island, but believe that would imply it was in mid-air. Splitting the pool into two halves would have no physical significance..

A weird feature is that the transport and cycling layers that, fail to render water here, do render it in the Trafalgar Square fountains, which use essentially the same construct (except that he outer is tagged as fence).

On balance, I believe this changeset moves in the correct direction and the problems lie with the renderers, not the map.

At least around Carlton Avenue East's junction with Preston Road, building geometry is poor compared with what can be obtained with Bing.

Also existing point of interest mappings haven't been integrated.

Google Street View must not be used as a source for OSM.

The access restriction is already mapped on the bollard. There is no need for anything else, with the exception that it should have emergency=yes, rather than a note about emergency access.

Unless there is signage that indicates that the stub roads are restricted, they should not, themselves be restricted. In this case, even if there are signs, I think there is no need to map them, as no properly working router is going to route that way, and no one is likely to get prosecuted if they do go up to the barrier.

Conversely, as the bollard allows bicycles, the restriction should be a qualified one, and exclude bicycles (and emergency vehicles).

I don't think this footway is called shelter. It might be a footway that should be tagged as covered osm.wiki/Key:covered Alternatively, may be it is is a building with a roof and no sides (i.e. only at level 1).

Should not have a name!

Is this really a playground in the OSM sense? Does it really have swings, roundabouts, etc.?

Again, don't use the name for a description.

In this case, there are tags that allow you to specify the number of disabled bays.

These car parks are are almost certainly part of the school. Please add access=private to prevent them being suggested as open to the public.

Please do not use name to describe age limits. If you know the limits, please use min_age and max_sge osm.wiki/Key:min_age

If you don't know the limits, just add a note giving information you used to make this distinction, i.e. provide the underlying facts, not the guesses you made based on them.

Please do not use name= to describe the feature. If it doesn't have a name, don't include one.

Please don't use name for a description. If you don't know the name, leave it out.

Also, if you don't know the building type, use building=yes and include any further information as a note.

See osm.org/note/206249

These aren't public footpaths. The highest conceivable access rating is destination, but I think they are probably best considered private. As such, they are difficult to, legally, verify on the ground.

The whole estate is NHS land and not a public highway.

This effectively reverses a change I made. I believe it is incorrect to map a path on top of a pedestrian area. The paths were already linked by the pedestrian area, both on the map, and in real life. If routing software cannot cope with that, the routing software is at fault.