[Tagging] How to put a name tag on an area with more than one type?

[Ture Pålsson]

2020-12-15 23:38 skrev Martin Koppenhoefer:

> Take a look back what I mentioned 3 days ago in my first answer: "...If we want to map all those "meta areas" with names we would do well to think about additional ways of delimiting space (i.e. different kind of geometry objects), e.g. a fuzzy border could be represented by providing different points for which it seems undisputed that they are in or out of the area in question. This would be very lightweight for all mappers, because it avoids clear lines which are confusing when they do not correspond with something observable."

This reminds me of my friend's somewhat tongue-in-cheek suggestion,
which I mentioned in another thread, that a fuzzy area should be mapped
as a function that takes a point as input and returns a number in [0,
1]; 0 for "definitely outside", 1 for "definitely inside". 

Here's an off-the-charts complicated idea for doing that, just to get us
started: 

A fuzzy area is mapped as a relation, type=fuzzy_area. Some tag on that
relation ('code', perhaps) contains a program written in a small
stack-based language offering geometric operators, such as distance,
intersects, touches, etc. The program will get a point as input on the
top of the stack, and should consume it and leave a number on top of the
stack. Geometric constants needed by the program (points, lines, areas)
are supplied as members (nodes, ways, multipolygon relations) of the
relation. If the 'code' tag is missing, all members should be nodes, and
the result will be computed using a default algorithm (which remains to
be defined) using those points. 

1/2 :-)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20201216/3dbbb267/attachment-0001.htm>