[Talk-GB] addr:place cleanup process

[Robert Whittaker (OSM lists)]

On Fri, 14 Jan 2022 at 19:10, Rob Nickerson <rob.j.nickerson at gmail.com> wrote:
> I think you are getting mixed up here. I suspect it is due to the so-called "Duck Tagging" test [1]. Just to reiterate the points from earlier in the address discussion, the settlement names in addr:hamlet, addr:village, addr:town and addr:city are wrong and should be ignored. The original tag (addr:city) should be used regardless of whether it is a town or city. As in, this would have been addr:postal_town if the address schema originated in the UK rather than in Germany.
>
> The addr:hamlet, addr:village, addr:town exist mainly because people have misunderstood this point above.

I don't think that's right. The wiki page
https://wiki.openstreetmap.org/wiki/Key:addr:suburb talks about using
addr:suburb for places *in the same settlement*, which suggests it was
only intended to be used for regions that are part of the larger
settlement. This fits precisely with the natural English definition of
a "suburb". As for the other tags, addr:hamlet is documented in the
wiki. While addr:village and addr:town are not documented, they both
have significant use in the UK, and addr:village has a reasonable
amount of use internationally too. I don't think these are really
errors from misunderstanding of addr:suburb, but a case of users not
having an appropriate documented tag to use for those elements in an
address, and so using what seemed natural.

I don't see a problem with using these settlement-based tags. They
wouldn't be strictly UK-specific as they're making use of already
defined place=* values, which are used internationally. I think it's
better to capture the information as to what sort of place we have in
the address, and it also solves the problem (that using addr:suburb
universally creates) of when we have more than one dependent locality
in an address hierarchy. In terms of addressing, I think it's
important to be able to distinguish between a "suburb" that's a
sub-part of the parent settlement, and a separate settlement that's
nearby but not part of the parent settlement.

With the exception of addr:city (which is set to correspond to the
post town, as defined in the wiki) I'd suggest we allow other
settlement based tags to be used, and define them to correspond to the
 accepted place=* values, which are already understood
internationally.

> If you have an address which has some settlement name before the postal town then there are only ever two tags that you need: addr:suburb and addr:city. The case we haven't yet worked out how to handle is when the address includes 3 settlements / settlement sub areas ("locality elements" in RM language). At that point we use addr:suburb for the smallest, addr:city for the largest and an as yet undefined tag for the middle.

If we do accept the use of addr:suburb for distinct places, then I
don't think this is the way to go when there are two extra levels. I
think you should use addr:suburb for the place immediately below
addr:city, and have a new tag for any smaller settlements below that.
This ensures that all the addresses including the settlement
immediately below addr:city (including those that are in that
settlement itself) will use the same addr:suburb value for that
settlement, which makes validation much easier. Otherwise, you'd end
up with two different addr:* keys both containing the same settlement
name. However, this might get rather messy if you had a suburb within
a town that isn't a postal district. Which is perhaps another reason
to allow the natural settlement names to be used in addr:*.

Robert.

-- 
Robert Whittaker