[Imports] TMC LCL - automatic script

[Frederik Ramm]

Hi,

Pieren wrote:
> It's not "discouraged" or only in Germany. 

I didn't say "only in Germany". I said that "as far as I am concerned", 
meaning that whenever someone asks me whether they should use 
type=boundary then my answer is no.

The reason for this is that multipolygon relations are something quit 
special, with special code having to be used to piece together the outer 
and inner rings and create proper polygons when evaluating the data. 
Since I want borders to use the very same mechanisms (e.g. allow any 
number of ways together forming the outer ring, allow enclaves/exclaves 
etc.), I am advocating the use of type=multipolygon for these too.

I fear that otherwise we'll have more and more relation types that want 
to use the multipolygon logic without being tagged a multipolygon, and 
that means that we'd have to change our software all the time. Instead, 
I say: Whenever you want to use this logic wher eouter rings and inner 
rings are pieced together from individual ways (or even sub-relations), 
please tag type=multipolygon so any software dealing with it knows what 
to do.

Andrzej is right in saying that a border is not a polygon but just a 
line (or a ring). However my view is that the border relation in fact 
represents the area of the country or administrative unit - which is 
*drawn* as an outline but conceivably, on a political map of the world, 
could also be drawn as a filled polygon.

I won't ask people go out and re-tag everything that is type=boundary 
but if you create new border polygons I think it is a good idea to use 
type=multipolygon and will continue to advocate this.

Bye
Frederik