> Gesendet: Dienstag, 16. Januar 2018 um 15:55 Uhr
> Von: "Marc Gemis" <marc.ge...@gmail.com>
> An: "Tag discussion, strategy and related tools" <tagging@openstreetmap.org>
> Betreff: Re: [Tagging] Sidewalks and cycleways as tags vs as extra lines
>
> adding e.g. surface=paving_stones to the cycleway and
> cycleway=paving_stones to the main road is tagging the same feature
> twice, not ?
Yes and no. You could deduct bits of information to complete the (tag) set
of information on anyone of both objects using the other respectively.
The decision to draw two separate ways to model an aspect of reality that
may also be abstracted/viewed as a single object (composing of..) is a pre-
selection on the understanding of that feature (with reasons).
But the reasons chosen are (to some extent) arbitrary, e.g. you could model
the kerb as a separate way and you will find that "one feature, one element"
justifies such a decision as well. Contrary, the assembly of road, lanes,
markings, sidewalks, manholes may be /viewed/ as a single feature, and this
abstraction in turn supports _not_ having separate geometries for each of
the roads components.
OSM models the world spatially, which means it seeks (questions to find)
boundaries to tell apart one feature from another. Some objects in real
world make this easy and intuitive (mostly man_made features), others
less so (some natural features).
Another aspect of OSM is the classification of similar objects with tags,
but this is (just as finding boundaries) another method of abstraction.
And there is necessarily a feedback between both of these methods.
One reason OSM models cycleways and sidewalks separately, is because it
is easier to classify (i.e. match properties and find similarities to
others) a road ensemble looking at its components. I.e. taking cycle-
ways and sidewalks out of the equation, fewer tag sets can be used to
decribe larger sets of roads with (almost) equal properties and it is
easier to identify a certain road to be similar to another.
There have been debates about whether to map lanes individually or not,
a lane may be viewed as a single feature or just as a component of a
road, so it is easy to see that this decision making (what boundaries
define "a" feature; in what respect may that feature be put into a
class with similar features) repeats fractal-like.
To come back to your original question, you would tag the same aspect
of reality twice, but in two different models of reality:
* If additional tags on (the centerline of the) road describe alongside
features, then effectively boundaries have been chosen that reach out
to include these alongside features within the feature itself. The
osm element models "a" a subject feature that is an ensemble of feat-
ures.
* The second model uses different ("narrower") boundaries, it describes
a sidewalk or cycleway area feature, neglecting alongside features (it
does not have tags informing about the presence or surface state about
the neighbouring road feature).
We can conclude that both models overlap in their description of reality,
but not to the same degree.
Greetings
cm
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging
