[OSM-talk] Mappers and apps should focus on relations at the very start

Tobias Knerr osm at tobias-knerr.de
Sat Jun 27 21:28:00 UTC 2015


On 27.06.2015 18:58, Fernando Trebien wrote:
> When this notion of "grouping" is
> presented at the very beginning, I believe people will easily
> understand it for all of the advanced scenarios

The notion of "grouping" would at least be more intuitive the common
"relationship between things", which is very remote from how relations
are often used. Still, it isn't really a natural fit for some relation
types, e.g. building, bridge, tunnel, ... relations.

So if we are discussing to re-frame relations in a more understandable
fashion, I wonder if we shouldn't turn relations into *objects*.

>From the current OSM environment, it seems clear that preset-based user
interfaces are popular. Mappers tend to grasp the concept of "creating
an object, then adding more information to it" quite easily. So what if
we base the data model around this?

Imagine mapping a building. Currently, there is a stark contrast between
the outline and the height of the building: One is a tag, and the other
is a way, i.e. geometry. Values can be many things - text, distances,
speed limits - but they cannot be geometry.

If the building was an "object", then both the height and the outline
could be added as values of tags. Naturally, there could also be
multiple keys with geometry values, resulting in the same expressive
power as our relations - but in a much less scary wrapping.

Of course, behind the scenes this would still look like a relation with
tags and members. It's all in the presentation and ubiquity. I believe
(a hypothesis that would be interesting to test) that this would make
the more complex features of OSM significantly more accessible.

So should we target this for API 0.7? Not really. I'm under no illusion
that something like this could happen anytime soon. But imo it's an
fascinating idea to think about.

Tobias



More information about the talk mailing list