[Talk-us-massachusetts] Plymouth Import Test

[Greg Troxel]

Greg Troxel <gdt at lexort.com> writes:

> I think that we should not be adding points with just street/number, but
> also town and state.   This is just making it harder to fix later since
> town and state belong.   I don't understand why you think it is lower
> risk to omit town state; that doesn't make it more likely that the other
> data is right.

The first question is what OSM norms are, in terms of if address points
should have town/state/etc., or whether they should not, or if it is
truly ok either way.  I have always had the impression that at least
town/state should be filled in, and it seems we've generally been doing
this when hand mapping.

A related question is if data consumers are obligated to process address
points with implicit town/state/country from level 8/4/2 admin
boundaries, and whether substantially all of them do.

A difficult issue is places like Barnstable (the level8 town), which as
I understand it has level9 (or 10?) divisions with names like Yarmouth,
Hyannis and even Barnstable, and in addressing these are treated as the
town name, even though the actual legal town is Barnstable.  I expect
there are more places like this, but I'm not sure of any specific ones.
(This is different than named neighborhoods, in that addresses use the
sub-names.)  I don't see how this can work with implicit town names.
But I have been saying repeatedly that we should leave Barnstable out
for now!

It could be that it's ok to omit town in the osm data, when the town in
the MAD address point matches the admin level8 name and thus the
implicit matching data consumers do is correct, and that we'll need to
be adding it for Barnstable .

Do we know that every point in MAD with the name Plymouth is within the
osm level8 polygon, and that every point in MAD within that polygon says
Plymouth?