[Talk-ca] PEI NRN import 21I16 - and questions about ways

[Tyler Gunn]

On Fri, 7 May 2010 08:19:43 -0300, Robert Shand <bob at shand.org.uk> wrote:
> However if I try to combine the ways I'm getting a conflict on the
geobase::uuid

You'll notice when you combine the segments into a single way that the
suggested uuid is basically all of the UUIDs for each segment separated by
a semicolon.
So you STILL do have knowledge of what Geobase data went into making up
the combined way.  

Granted you do lose the ability to split the way up and know what parts of
it correspond to which UUID.  However I don't know that this is a problem. 
As other OSM users edit the data and refine it in their own ways, these
ways could be split, combined, etc.  I don't think its reasonable to assume
that the Geobase data will be imported and left in the EXACT format it was
put in.

My vote would be to combine the ways as you've suggested and accept the
semicolon separated UUID list.  

I've already run into this a few times when creating address interpolation
ways with the StatsCan data and I've used your approach.

Tyler