[OSM-dev] AND data integration

[Jeroen Dekkers]

At Wed, 18 Jul 2007 16:05:33 +0200 (CEST),
Sander Hoentjen wrote:
> possibility 3) Like 2, only now we write awesome scripts that detect
> duplicate ways, only import data not existing yet and updating existing
> ways with extra information that is available
> - Hard to do, expect to finish at T=100 to infinity, depending on how
> perfect the script must be (there is no such thing as perfect).
> - What is left is to fix up manually the few things the scripts couldn't,
> totally done at > T=120
> downsides: only technically skilled people can contribute and they are a
> minority, it takes long, hard to decide when the scripts are "good enough"
> to finally do the import
> - upsides: if done properly, the scripts can be reused when new data is
> given to us, no existing work is lost

I think this not that hard to do and that it doesn't have to take
longer than the other options. It should be quite easy to see whether
a street is already there by matching the names. I guess you can also
just pick a way and match it to the closest way in the other data
set. It would take some time to write the code, but I don't think it's
that hard.

IMHO this is the best approach, the only reason not do go this route
is when there aren't enough technically skilled people to write this
stuff and there is a big group of people willing to do the manual
merge. Personally I would really like to help develop a merge tool,
but I simply don't have the time at the moment. So I'm wondering how
many people can do this and do have the time?

But even before we can start merging we first need to make sure we
know how to convert the AND data to the OSM data model without losing
information. I don't think we are anywhere near that yet. It looks
that people are mostly working with the roads. I haven't seen anyone
working converting the water boundaries for example.

One big thing that we need to think about is how to implement
navigation things like turn restrictions in OSM. We currently don't
have those features yet.

Other than that, there are some people who are already programming
things, but I haven't seen much code yet, so it's difficult to judge
what has already been done and it's also difficult to avoid
duplicating work. So it might be a good idea to make a directory in
SVN (something like applications/utils/and) and put all the code in
there.

Jeroen Dekkers