[Talk-us-massachusetts] Want to test Phase 0 MAD import in Plymouth

[Greg Troxel]

"Alan & Ruth Bragg" <alan.ruth.bragg at gmail.com> writes:

> I've got a Plymouth OSM user who's willing to help me validate an import.
> He's trying to add addresses to support the town's 400th anniversary this
> year. He's seen the Plymouth Street name conflicts and is willing to
> provide local knowledge to change OSM after the import.
> I plan to use my import account to officially test a phase 0 import as
> described in the Wiki.
> I'll notify the MA list and point out that the whole thing can be reverted
> if necessary.
>
> I'd like to have your concurrence or suggestions before proceeding.

My concerns are basically that there are a lot of consistency checks
that I think ought to be done, but which are only sort or described in
the wiki page, and I don't think we have any code support for.  I've
sent them to the list, and tried to edit them in, but the page is (or
was) disjointed as I put in normative text and other test describes
procedures that I don't think do what the other text says.

The normal import notion is to only import correct data, not to import
things that are 80% right and fix them up.  I don't know what % we would
be at and what the threshold is, but I'm guessing it's at least high
90s; 90% isn't good enough and 99.9% seems like it is -- it is a tough
call where to draw the line.  I would like us to be sufficiently careful
that the International Committee for Complaining About Imports has no
grounds to complain.

Is what you are proposing to do, going from a file from MassGIS, to an
osc to upload, going to verify that each address being added has a town
that matches the town that the building centroid is in, and a street
that matches a street in that town, that is one of the few closest
streets, that the building the tags are being put on has no existing
tags, that there is only one address being added to each building, etc.?
(I know this is hard.)

Do you know how many addresses we are talking about, and how those sort
into a single address on a building, vs more complicated things?

It might be that limiting this to single addresses on single buildings
would avoid a lot of issues; I'm not sure if that's part of your phase 0
notion.  It seems to me that multiple addresses and particularly
multiple unit numbers are harder to say what is right, and have a higher
potential to require cleanup work.

I think it probably would be a good thing to prepare the osc file or
files that you would propose to upload for review.  Then people can look
and have a less theoretical discussion.

(I still intend to dig into this, but I am still not getting to it.)