[OSM-dev] [Imports] Inserting OSM data

[Andy Allan]

On Mon, Mar 22, 2010 at 7:12 PM, Ian Dees <ian.dees at gmail.com> wrote:

> Almost 100% of the time these are imported to allow for the possibility of
> future updates to the existing imports.

Except, as you point out, they can't be used in any way for "future
updates" since you've got no idea if the reference stays on the
correct object. So I'd rather spend time explaining to people that
external references are a waste of time than encouraging them any
further.

> In order to have this possibility,
> there needs to be a consistent/persistent identifier for a particular
> "thing" on the map. Since we don't have support for that in OSM this is the
> next best thing.

No, it's a distraction from the real problem, which is that merging
datasets is fundamentally hard, and needs to be approached with
sophistication and fuzzy location-based matching (e.g. feature church,
near this point, name something close to "all saint(')s"). Slapping
external IDs on OSM data and crossing fingers isn't helping.

> Anyway, I'm not advocating for or against anything, simply explaining what
> I've experienced as an importer and someone helping others import.

Has anyone experienced sync working? I know a lot of people have
imported with external IDs but I suspect that it's not going to work
when they actually come to updating one way or the other.

Cheers,
Andy