This is obviously wrong when TKO is more than that.

1. `place=town` is an abstract concept that shouldn't have a defined area
2. New Town or OZP is for planning purpose only, and is copyrighted.
3. It is debatable whether they are `boundary=administartive`. Postal address is a different matter.

Moved to `name:yue-Latn=`

Moved to `name:yue-Latn=`

5. Please don't make up English name by transliteration or romanization. `name:*=` is for those in common use by the public. Use `old_name:yue-Latn`.

There's no `natural=pass`. In OSM. `=saddle` and `mountain_pass=yes` are distinguished clearly. The former is a low point compared to the surroundings on a mountain. The latter is the highest point of a road going through a mountain. There can be a mountain pass without saddle, and vice versa. They often do not coincident.

3. Please use more descriptive changeset comments, not repeating the same vague sentence
4. Please group similar changes into one changeset to be uploaded. This is excessive fragmentation flooding the feed.

1. Please don't cite random "online" source. `name=` is the common name in use by the public currently, not what you or someone or history think it should be.
2. This was considered an `old_name=` by previous discussions. Please raise this issue again.

It's not needed to have address-only names, unless it has special styling. (eg 700 Nathan; Nathan 700, Seven Hundred Nathan, etc)

Ok, I think you two are thinking about different parts. Let me see....

Please don't add descriptive labels to `name=`. That's for actual proper names only. Use `description:zh=`.

What's wrong with this one? There's clearly something there.

1. Please don't cite random "online" source. `name=` is the common name in use by the public currently, not what you or someone or history think it should be.
2. This was considered an `old_name=` by previous discussions. Please raise this issue again.

Please don't add `ele=` to `=ridge`. This doesn't make sense.

1. Another example is Chek Lap Kok being included in Tung Chung

(Btw if you are using your own instance, you could adjust how it works)

1. This would be a problem on Nominatm. It should return the island's name when there is one, as it is important as an address. However that's not what they do, so I don't understand what's their logic. https://github.com/osm-search/Nominatim/issues/1746 But it would be exhaustive and redundant to add a `=town` to every `=island`. Here, the important "Lantau Island" isn't listed for addresses on it either. That needs to be fixed first. This might help contain Discovery Bay within it. Unfortunately, the situation is Lantau Island is splitted between 3 District Council districts. I need to check how to handle this.
2. The problem here is there is no `boundary=administrative` below `admin_level=6 `anywhere, which in reality isn't even used for addressing. But adding `place=town` around might not solve it. On the contrary. you might cause Peng Chau to be returned on Lantau Island around the triangle intersection with Discovery Bay and Mui Wo, if Tai Shui Hang is downgraded from `=village` ( it should probably be a `=hamlet`) Discovery Bay is already returned for Disneyland in Penny's Bay, and the north shore on the other side of the mountain. That's clearly wrong too.
3. You did say "towns on islands". I need to review the definition for major populated islands here.

Please understand that it is considered a `place=island`, not `=town`. relation/9279538

125896795,125899891,125899201,125899117,125897623,125897588,125897548,125897332,125896912,125896876,125896822,125896795,125896578,125896523,125895298,125895099,125895015,125894946,125894883,125894042,125893997,125893643

125896795.125899891.125899201.125899117.125897623.125897588.125897548.125897332.125896912.125896876.125896822.125896795.125896578.125896523.125895298.125895099.125895015.125894946.125894883.125894042.125893997.125893643