[Tagging] Subtag for place=locality?

[Joseph Eisenberg]

I added some comments to the talk page of your "type=group" relation proposal.

I would recommend simplifying the proposal to just be for groups of
nodes, because there are already relations for multipolygons (areas)
and linear ways (waterway, route, etc), and it will be very difficulty
for database users to handle multiple types of objects in one
relation.

Also, I believe every database object needs a feature tag.
Multipolygon relations work in this way too. This may mean creating
new tags specific to groups of objects in some cases, but it's not
good to depend on taking tags off of the members of the relation,
since this will often lead to invalid results and is not currently
supported by editors like ID and JOSM, or most database users.

On 4/16/19, Martin Koppenhoefer <dieterdreist at gmail.com> wrote:
>
>
> sent from a phone
>
>> On 16. Apr 2019, at 09:13, Joseph Eisenberg <joseph.eisenberg at gmail.com>
>> wrote:
>>
>> A "group" usually has more than 2 members, but I can't think of an
>> objective cut-off point above 2 or 3. If "three's a crowd" it's also a
>> group, no?
>>
>> So I think it's reasonable for mappers using place=archipelago to
>> describe a group of as few as 2 or 3 islands.
>>
>> Similarly, if you use a tag like "natural=lake_group" to describe
>> named groups of interconnected lakes, this could be used for as few as
>> 2 or 3 lakes, or as many as hundreds.
>
>
> the group relation requires at least 2 objects:
> https://wiki.openstreetmap.org/wiki/Proposed_features/Group_Relation
>
> It may not be absolutely typical natural language to speak of a group of
> two, but proposing also a pair relation seemed overkill ;-) and a group of
> three is fine.
>
> You would not need to specify whether it is a group of trees, lakes or
> islands, because it should be implicit through the members, but of course
> you could add tags like natural=archipelago if it makes sense to you.
>
>
> Cheers, Martin
>
>