Am Fr., 7. Feb. 2020 um 11:03 Uhr schrieb Joseph Eisenberg <
joseph.eisenb...@gmail.com>:

> 1) The tag `area=yes` is only supposed to mean "this closed way is an
> area, not a line", and is only used when this is not already obvious
> from other tags.
>
> It is not necessary to add `area=yes` when the closed way already has
> a tag which defines an area, such as `landuse=forest` or
> `natural=scrub` or `leisure=playground`: all of these keys are always
> areas when mapped as closed ways, so they do not require `area=yes`,
> whether or not `barrier=hedge` or `barrier=fence` is added.
>
> This also means that `area=yes` is never needed when the same object
> is turned into a multipolygon relation, so checking for `area=yes`
> alone is not sufficient
>
> A closed way with `hedge=barrier` + `natural=scrub` is an area, and
> the presence or abscense of `area=yes` should not be required. This
> makes this a poor tag to rely on for rendering decisions, at least for
> a style that influences how mappers use tags.
>


in our data model, some area-feature plus barrier=hedge can mean: a hedge
(linear) around the area. Actually, it is what people do, it is not really
something that was conceptionally designed (contrarily, it is against the
one-feature-one-element rule but "we" have decided to support this kind of
mapping rather than for example ignore the whole object as "invalid"). If
we could educate people (through rendering decisions) to choose
representations that get rid of these ambiguities, it could be beneficial
in the long run. Rejecting all hedges mapped as areas by ignoring the area
tag for rendering, doesn't seem a sensible response. It is a 180 degree
turn, from supporting this style for hedges mapped as area to making it
impossible to apple the barrier=hedge tag on areas in a way that it gets
rendered.



>
> 2) Many hedges which were mapped like areas are currently missing
> `area=yes` tags. In this comment
> (
> https://github.com/gravitystorm/openstreetmap-carto/pull/3844#issuecomment-582692389
> )
> you can see that over 90% of the `barrier=hedge` closed ways in a
> Dutch province (random example) are missing `area=yes`



I do not have much knowledge about the Dutch map, but I recall they have
imported their landuse once and at least one time updated it with a
following import. Maybe these missing tags are missing because they were
forgotten in the import?



> , though they
> appear to be mapping the outline/area of the hedge. This means that a
> rendering solution that relies on `area=yes` would miss a large
> percentage of hedges mapped in this way.



these have been broken before and would be broken afterwards as well, no?



>
> Most likely, many mappers do not see why they should need to add a tag
> like `area=yes` when they map a hedge as a closed way.
>


indeed, they do not need, only if they intend to represent the whole
enclosed area rather than a hedge as perimeter they would have to add
area=yes.
Clearly, it isn't nice to have a tag for something basic as the distinction
between the line and polygon geometry type, and maybe this will be solved
more generally in the future by adding an additional geometry type.


>
> A better solution is to use different tags for linear features and
> area features. This is the standard used in Openstreetmap for common
> linear features such as paths, roads, railways and waterways: the area
> and the line are different tags.
>


interestingly, for paths and roads there is also an area=yes variant (which
is likely more common than the newer "area:highway" tag, which has
different semantics). For example there are 158502 (27% of all pedestrian
objects) instances of highway=pedestrian, area=yes alone. It's the
standard. ;-)




>
> (The rare exception to the rule that the same tag isn't used for lines
> and areas is for pedestrian plazas which are mapped with area=yes, but
> in this case the idea is that there is no linear feature: the area
> allows travel in "any" direction by pedestrians, so there is no "line"
> to map, just the area.



exactly, just like the case of barrier=hedge as an area.



> Yet there often is confusion about how
> pedestrian highway areas should be mapped and how they differ from
> area:highway=, so this cannot be considered a good example to emulate
> in the future. )



IMHO these are sign of a lack of awareness of squares as important feature
in the urban structure amongst the early contributors (the idea and focus
was the creation of a routing graph). Later on, really late, place=square
was added.
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Reply via email to