[OSM-talk] Segments, ways and order

Matthias Julius lists at julius-net.net
Fri Jul 20 13:59:11 BST 2007

Frederik Ramm <frederik at remote.org> writes:

>> The biggest challenge would probably be to automatically convert as
>> much data as possible.
> In fact, it is quite obvious where the problems will lie once we start
> this conversion: branching ways, and non-contiguous ways.
> Both of these can be identified and, mostly, changed to simple
> structures *now*, without changing the data model one bit, so that
> automatic conversion will be easier once it is attempted.
> I am already doing this now; whenever I encounter such ways, I change
> them into simple linear structures.

Me too.  Especially since I installed the validator plugin which
complains about those.

> I'm also experimenting with a Perl script to find those cases that
> cannot be converted automatically. (Those that are auto-convertable
> are not too interesting for now.)

One area where non-contiguous ways are regularly used are rivers and
lakes with islands.  Those would be easier if you introduce
superways/advanced relationships at the same time.


More information about the talk mailing list