No, I am not sure it is just called "Public Transport Interchange".

At least we can't have a roof that holds the name of the PTI. That's already an improvement. Actual name will be left as a future todo.

passing by

way/853595683 looks like should be `oneway=yes`

Terrain issues addressed in changeset/180747693

Building level issues, the thing is that the validator kept complaining about road-building intersection. I had to address that somehow, and this should be best reviewed in a follow-up changeset.

OK because you are interested in this...

-----

You may notice this junction shape keeps reappearing throughout Tin Shui Wai. Another example is e.g. way/1354166352 . Granted, that was a long time ago, so the specs back then may not match what we expect today.

Still, because the road does widen symmetrically at this place and the left turn therefore diverges away from the median, I thought it appropriate to draw the split early. Drivers are allowed to enter the turn area and instantly decide to go back to the central area, but this maneuver is unsafe and makes no sense so imo can be safely ignored.

I have never heard of any frontage road that exists not on the far side, but on the opposite side of a road. It may not be fully accurate to classify them as "frontage road" but hopefully you get my point. The examples here do not collect traffic back to the main westbound PER; they are "internal only".

Not slippery. At minimum, street names must match to have a chance to be "=*_link". Here, it is still called "Prince Edward Road West", it is a component of PER (currently "=primary") and we can give it "=*_link".

imo the segment does not feel like "internal road", because it brings traffic from outside the block to inside it. (True "internal roads" see east side frontage road of PER.) But if it is "internal road" then yeah do "=unclassified" or below.

Well, looking at the OSM history, safe to say it shouldn't be "tertiary" (first version), and shouldn't be "driveway" (my mistake).

The idea is, because Nullah Road is forced to enter PER, this can be considered a link away from PER so to reach Tung Choi Street. It just so happens to also be a driveway. Also see small section of Fa Yuen Street also being a primary_link.

Socrates method. I point towards the extremes and then hopefully can deduce The Real Thing.

Re "remove inside lines", supposedly routers should treat pedestrian areas as routable even when there are no foot paths inside of them, but the exact implementations are specific to each router and very ambiguous (eg what if we already provided some paths?) so it's getting nowhere. But routers should know how to deal with this kind of shape. At minimum it just routes along the outer edge as if it's a circular road.

With both examples, I kinda see where you are coming from, and I can see the issue being the wording of "omnidirectional routing". The quoted pedestrian area in Ma On Shan is square shaped but realistically only has two exits that sits along its axis, so it doesn't seem omnidirectional. But this should be sent to the forums for further clarification.

I can just remake this as area:highway for now.

For inspiration, how about same district case way/620136042 ? Most navigation would occur along the bridge axis; does that count as "omnidirectional"?

Hi there, we noticed node/13349030768 probably has mistaken info. May you take a second look?

I don't think those are tree rows. I think those are general areas where trees are planted to grow / may naturally grow.

I passed by this area some time before. Notice how there is now only 4 poles, yet hkbusfandom describes 5 (old info). I did not pay too much attention to the refs, but if the poles are relocated, we have reason to believe the refs may be different.

I see we are only disagreeing on whether this counts as "mostly linear"; if this is indeed "mostly linear" then yes we should do `area:highway=*`. I will just make a new note to collect more opinion on this.

I was trying to see how GraphHopper performs when I ask about a path going through Route 7, but I later remembered GraphHopper by default no longer snaps to bridges so it seemingly cannot navigate to/through Pung Loi Road.

Re the quoted relation, see relation/1807461/history/2

This implies we can/may add `maxspeed:type` etc to all the roads in Hong Kong. That's gonna be too much work. At least I will just stay out of this kind of implicit `maxspeed=*` in the future.

The goal is to fix e.g. GraphHopper not able to enter these roads. A common reason for this is confusing turn restrictions that make some roads "impossible" to navigate to.

I get what you are trying to say, but I don't think this is a case of "almost linear". We can't reduce the geometry here to a straight line because there would be a triangular path in the area, and then the mapping would be inconcise.

Consider peds that enter from Kam Chun Court. They either go straight for the bus stop (I think you mean this when say "linear"), or they rotate and approach the footbridge. This creates some sort of "desire path" that diverges from the linear structure.

There is also a potential problem/concern that a hypothetical centerline footpath + `area:highway=yes` would be too far away from the bus stop nodes and would make routers and GPS confused (would snap to vehicular lanes instead). The depth of the area is 14 meters, which is way higher than existing GPS resolution of 5 meters.

Upon review, we do also have extra (outdated) tags on way/1178516528 that should be removed.

`area:highway=` implies "non-routable" but that's not the case here for way/1486486342 . While most routers treat pedestrian areas as a circular road using their outer edges (e.g. GraphHopper), it's very coincidentally convenient that the outer edge of such area is exactly how people would navigate through this pedestrian area.

Conversely, let's say hypothetically these roads never had any maxspeed info ever since OSM was founded. It still wouldn't make sense to suddenly add maxspeed to them.

"Adding 50km/h everywhere" should be used sparingly as band-aid fixes when in the past the various routers can't understand speed correctly, but now is not the past, we (nearly) have the tech and are improving it, we just don't need to do that anymore.

Chill.

These local-only roads do not have any irl maxspeed signs. The maxspeed aspect is better handled/described by country rules.

You may notice that I basically only add maxspeed to explicitly-signed roads and "roads that look like proper highways (eg Ma On Shan Bypass)". I also extend these maxspeed information towards the next junction down the road, which should be uncontroversial.

This is not the kind of "haha delet this" random action that is often done by newcomers.