Good catch. Improved / fixed with changeset/179170787

My understanding of emergency access points is that they only have to be "close enough" to the actual rails. For example, the Chinese HSR Kwai Chung EAP (not in OSM, currently transitional housing) is not on top of the rails, but actually beside them.

I vaguely remember there will be 3 tall residential buildings for the Developments, 1 on the west plot and 2 on the east plot (citation needed), but basically, surface level works (e.g. piping) can't go that deep, and only the tall buildings will interact with the railway tunnels. Imagine/Expect a few years later when satellite imagery finally shows how the new buildings are actually placed, and then we can further review the railway curves.

To reproduce/verify this changeset, on the surface level you can count the number of left/right turns. For a deeper look, you can measure the rail curvature by standing at the car links and then noting how your chord intersects the arc.

More details can be given at Discord to avoid cluttering this for too much, but also because there may be image explanations.

Several observations on past data:

- if segment at Chuk Yuen Estate is full straight, then why no extra station? therefore, not straight.
- railways do avoid e.g. Hsin Kuang Centre, but the fit is still too tight.
- by survey, the left/right turns in opposite directions do not match each other; therefore, obviously asymmetric.

made a new note note/5177281

Good question on Nullah Road -> Tung Choi Street; at first glance this is dangerous and likely not a popular/viable path.

Then, need to confirm whether lane markings allow this (half-think cannot).

Ultimately, driveways is a subset of service roads, for single-property access, compared to service roads that may form "technical" roads inside a single facility, or somehow linking several adjacent related facilities. I imagine the quoted way should be described as a driveway instead of a generic service road.

Or, let's put it this way:

This segment only leads to the police station, and only the police station needs this segment. No one uses this segment for through traffic (see Tung Choi Street some distances before).

Therefore, it definitely is a driveway, and not a frontage road. The "frontage road" rule doesn't apply.

That's clearly a driveway for the police station, and not a frontage road. Notice how it is right outside of the reporting room.

tbh there is no gyratory; just use the turn lanes etc and then vehicles could get where they need to go, no need to drive into Boundary Street.

The physical shape is basically asking for smooth merging from Lam Tei Interchange (ground) to CPR, while giving priority to Route 9 exit. Just that, it's not feasible to draw a yield line along the travel direction (way too long), so they just went with a yield line right at where the lane separation ends.

It did momentarily occur to me "I should do undelete", but then I thought, "the original straight-only is consequential because the real one should be no-uturn", so I just made a new one.

Also iD clicking is easier than JOSM clicking.

Note:

I tried to update the bus relations where reasonable, but it turns out some bus relations are just plain broken and should probably be reviewed by some public transport mappers.

Again, I always expect Mapillary to have no data at all, at least for Hong Kong.

To clarify, I know "straight" junctions should keep straight. What you may notice about "keep straight" may be because the slip path is short so that the "keep angle" rule forces the short path to be straight all the way (I personally find them ugly).

However what seems unspecified is this kind of "far side slip" junctions.

Ah, good find. Usually turning right does imply "u-turn is allowed".

Will make a todo note for now. The plan is to somehow add e.g. source = mapillary to the turn restriction itself so in the future people will know something is up when they see a source tag.

I have heard about something like this before, but to me it feels more like some sort of artistic discussion. This would be different from e.g. earlier discussed "dashed lines separation".

And, specifically, for this case, I did not know / did not notice where the control box is at.

... ????????

iirc I have never declared my stance on the mapping style of "free ref node"; this is observed to be seemingly a convention.

@kingkingHK You quoted a changeset. To clarify, my comment on that changeset is essentially "that free ref node shouldn't exist because there's no traffic signals at the roundabout". There is no judgement on whether it's good or bad.

At minimum only you @kingkingHK found the "free ref" mapping style to be of poor-quality. I find merits of that style, because now I can very easily know where a specific ref traffic signal is located.

@Kovoschiz mentioned putting the ref at the control box. This is definitely correct as expected from the style of OSM Global, but sometimes we just don't know where that box is, or maybe there are multiple boxes for the same set of signals.

It seems to me, by convention, there should have a single free node that contains the traffic signals ref info, and then several based nodes that actually describes where/how the traffic signals apply.

1. aerial imagery by ESRI and MapBox seem to show a triangle, which indicates a yield sign
2. I was about to say, need not worry too much since soon we will need to map Yuen Long South developments and there will likely be junction upgrades, but I see you have already remade the changeset

info: there is a South Asian (Pakistani?) community here. what probably happens is that during some hours, some of the folks would lay out some shops (usually at Pei Ho Street due to low traffic) and sell homemade stuff and chat among themselves. doesn't feel like a real tourist attraction.