Hi there, the irl signage is "no u-turn", and the meaning is "straight on to the refuse collection point entrance only".

irl there is a (I think always closed) roller shutter that acts as the "back door" of the refuse collection point located here.

The problem is "always closed".

I don't think OSM cares about shop entrances, but maybe I just don't know enough.

If people care enough then they can just look at this changeset and know this new node replaces 2 previous nodes. There is no unified history to speak of when features are erroneously duplicated.

My apologies for the "original research" accusation. Being unable to locate all the 4 related exist irl has made me very suspicious of the situation, and it turns out there are official documents backing up the previous observation.

The current MTR copy of the railway protection boundaries agrees with the Building Department's copy

https://www.mtr.com.hk/archive/corporate/en/operations/ISL_RPAP/Mtr_Ap_59.pdf

Turn curves would be style choices.

There was mismatch, as in "these bus bays from satellite imagery don't even exist in OSM data".

The following was considered:

- old paths did not fully represent "new path shape"; can't just flip them
- afaik all concrete covers were demolished; Theseus' Ship
- no absolute requirement to maintain path history, and sometimes it's just impossible (e.g. point bus terminus -> area bus terminus, an improvement but "loses history")
- all bus routes are reassigned new platforms; "keeping history" doesn't see data maintenance convenience
- these will eventually be fully demolished by phase anyway, as irl construction further progresses

As such, it was decided to fully redraw everything.

Hi there, do you have more information about the Quarry Bay B1, B2, B3 and B4 exits? These information seems like original research, and cannot be submitted to OSM.

I went there again, and yes I saw it wrongly. Those are supporting structural walls, not pillars.

This may suggest `tunnel=yes` but consider these tracks are actually at ground level (observe from Shek Kok Road), and one may see sunlight while riding TKL at this location. Doesn't make full sense to get `tunnel=yes` (eg contrast Yau Tong Station where sunlight does not exist).

Might we do `covered=yes`?

The tracks can see supporting pillars. Wan Po Road looks like `=bridge` to me.

I can extrapolate the discussion to cover "integrated bridges" e.g. Tai Chung Road in Tsuen Wan currently mapped as "at surface" but is technically a large bridge over a covered ditch.

Hi there, do you have more information about 多廟徑?

afaik OSM really strongly discourages using `=institutional` because it does not distinguish between e.g. "government", "medical", "education", "ngo" and a whole lot of other categories.

My goal on OSM is to provide/handle updates quick enough with reasonable correctness. Sometimes I choose not to ask when I expect the change to be simple / small, e.g. this case.

I do admit I was thinking quickly here and then selected the crematorium preset on the iD editor.

Improved with changeset/171031532

It turns out we don't even seem to have `landuse=amenity` in OSM.

The original changeset was to update on the `landuse=construction` tag. I guess we do need to remove the `landuse` area altogether.

I have a feeling the chief OSM schema guys mostly live in places where burial is done in the traditional sense of "graveyard" and have no concept of "a building for holding ashes".

I currently view `landuse=cemetery` to be "a land that is used to hold/bury dead bodies". This would work alongside e.g `landuse=residential` to be "a land that contains houses/homes for people to live in" without needing to specify the exact format of homes/houses.

But still, one thing to note: the distance between the tunnel openings on both sides really are different. I think there should be some unusual shapes here.

I also need to apologize here. Others e.g. Kovoschiz might already know it, but I tend to think/write/type directly, which I think has the tendency of scaring off people. Please don't be too scared 🥲.

Tbh I don't quite get the optimal usage of changeset comments. The thing is that the changeset author will be notified of new comments. I just take new comments as something I should respond. Maybe I got the entire feature wrong.

Regarding the sudden curve, I took a look at the ESRI imagery again, and again, you are right: the east side also is misaligned. Indeed, my intention was to "finish the changeset asap" and the curve is what I decided to do because I thought the east side was correct. But it was/is not. I have made a decision based on wrong info.

Thank you for pointing this out! I will write down a todo in the map so that me or others can improve this area in the future. I do spend less time on OSM nowadays compared to some time ago so some of my changesets perhaps is not the best.

Hi there!

Re aerial imagery source. It should be a community consensus to use the ESRI aerial imagery (not "ESRI Clarity beta"!) since it is often the latest (I think now is July 2025), and it is OK-ish clear to correctly see the features. This changeset was made with this ESRI imagery.

Re altitude distortion. Yes, your view is correct, and I also know about this.

Re quoted dubious calibration. Yes, it was also made by me (woah).

My objective in this changeset is to look at the imagery and describe the latest observed "new roads" from imagery. However, doing so requires alignment with satellite imagery, which existing roads do not align well. Some creative work is done to balance between aligning with satellite imagery and retaining reasonable curves with existing data.

Ultimately, if we really need actual GPS tools to correctly provide the coordinates (tools which I do not have), then I believe we may be requiring too much a barrier for OSM editing. And, as much as Sisyphus is concerned, GPS coordinates can self-drift by itself, and imagery can also self-drift by itself, so requiring absolute accuracy is too nihilistic.

IMO, a bigger context is that, there is a guideline that discourages using Bing imagery (see osm.wiki/Hong_Kong#%E7%8F%BE%E6%99%82%E8%A6%86%E8%93%8B%E7%8E%87_Current_coverage ), and when I see the roads distorting too much from imagery, I cannot help but think "perhaps this was first made with Bing", and then go edit them. There is a real chance the initial shape is just too bad, and my improvements are legitimate.

But for real, if you think the changeset was too bad, you can go improve upon it. Chances are I won't notice subsequent edits, and there will be no edit wars.

Re `=use_sidepath`, that is my negligence, and I will review it again.

Re `access=customers`, respectably disagree. imo `access=customers` implies there is some sort of (human) guard that prevents you from getting into the area (e.g. Shing Mun Tunnel BBI, the exit to Shing Mun Road is there but is guarded). Shing Mun Tunnel BBI should use `access=customers`, I agree with that. But here, and at several other BBIs, outsiders cannot enter the area not because there are any (human) guards, but because it is simply physically impossible to do so. There are no guards.

The network really is free-hanging, as correctly warned by the iD editor. Having `access=customers` here would then be an unusual kind of "consequential mapping", and should not be done.

I see no problems here.

Merge conflict with changeset/170379585 ; changeset/170379585 was used.

Review may be needed.