[OSM-talk] Request for Romano-British features

Martin Koppenhoefer dieterdreist at gmail.com
Sun Jan 15 00:41:57 GMT 2012


2012/1/14 Russ Nelson <nelson at crynwr.com>:
> Lester Caine writes:
>  > yet there seems to be no way to create secondary databases of
>  > information which can be used in parallel but separately hosted.
>
> Yes, we need *more* OpenStreetMaps from which you can fetch different
> types of information. For example, any data which is externally
> maintained and updated really *shouldn't* be imported into OSM.


If you really want to connect the data you will have to have it in one
database. Layers are only good for stuff that is not connected/linked,
i.e. that is independent from each other (rarely the case for features
like the ones we are talking about here, as existence of the roman
road had an influence on the further development of the road grid).
E.g. if you have a building of which the basement is roman and the
upper parts are medieval (we can asume for this example that they both
occupy the exactly same area), and you realign the upper parts (say
you slightly move 2 corners) you will also want the lower one to
follow this modification. The both are linked and for consistency a
refinement of one will (in many cases) require also the other to be
modified.

We would need to have a way to link 2 datasets like: this line in
dataset A is the same as this line in dataset B (or this line in A is
parallel to this line in B). If you continue to think this and think
about the complexity of the real world (there is not only a few lines
but many of them, and they are not only the same or parallel but also
orthogonal, or in the middle/center, or ...) you will see that you
either keep them in one dataset or you will loose all these relations.

Layers work well if both are only loosely linked (say you have
temperature measurements in a grid and overlay them with OSM data for
visualization).

cheers,
Martin



More information about the talk mailing list