Hi Clifford,

I don't support importing these id tags. For updates, the optimal
identity of an OSM object is a combination of its location, size, and
name/address.You don't have to believe me, Paul has perfectly working
code that does updates to OSM addresses without a primary db key.

https://github.com/pnorman/addressmerge

We keep them, because It is the same part of our primitive brains that
fills up our houses with stuff that we we never use and keeps 20 years
of email. It is the OSM manifestation of peoples aversion to loss
(http://en.wikipedia.org/wiki/Loss_aversion).

These extra tags do have real costs to onboarding new users. It was 6
months before I was comfortable touching the gnis nodes in my area
because of the id. New mappers will be less comfortable fixing bad
address when they see a tag that is linked back to the local
government GIS department. Right?

I wish Chicago, NY City, NHD, and GNIS were imported without the ID's.

To answer your question, to deal with the case that a single address
might have IDs from different external data sets, I don't think we can
have one id tag. The actual external source of the id needs to part of
the tag. <source>:id is fine.

Thanks
Jason.




On Sat, Dec 14, 2013 at 1:50 AM, Clifford Snow <[email protected]> wrote:
> We are starting work on the next phase of importing addresses for the rest
> of the (King) county. We plan to use the latest address data from the
> county. Partly inspired by the work being done if NYC, the county is now
> publishing an record id for use in future updates. We want to incorporate
> this field into the import. But there doesn't seem to be a standardized tag
> for this id. For example, GNIS is gnis:id, tiger has another scheme and I
> believe NHD has another id.
>
> Eric Fischer got me thinking when he was working on a tiger fix that we need
> to retain this data for future updates, even if we don't know how to do
> updates today. Rather than use kingcounty:id, shouldn't we have a consistent
> method of identifying tags that may be needed in the future? It's possible
> just using :id is sufficient. However, maybe something like update:id,
> although that has problems as well.
>
> Thoughts?
>
> --
> Clifford
>
> OpenStreetMap: Maps with a human touch
>
> _______________________________________________
> Imports-us mailing list
> [email protected]
> https://lists.openstreetmap.org/listinfo/imports-us
>

_______________________________________________
Imports-us mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports-us

Reply via email to