[OSM-talk] A post box called Breuningsweiler

[Sebastian Spaeth]

David Earl wrote:
>> No, you found a town called Friern Barnet, the centre of which
>> happens to be a
>> road junction that has traffic signals. The town is in Barnet, London,
>> England, United Kingdom.

>> If you are route planning, isn't it better to be able to route
>> directly to a
>> node with the name you are trying to get to, rather than route to
>> the nearest
>> node that is part of a highway that happens to be somewhere
>> adjacent to the
>> place you are trying to get to?

> Almost no one (i.e. all the other places except those I found) has done it
> this way. It's not usual practice, and therefore it is going to cause
> problems. Equally, no route planner can behave as you suggest because the
> majority of places aren't done this way (and what if I'm navigating to a
> pub, does that have to be conected with the highway too? And in any case
> navigating to a place doesn't mean ending up at some arbitrary spot that
> someone happens to have chosen to represent the place, which is a much
> larger area in practice).

Then I am also guilty of being another "almost no one". I try to tag
places at nodes that are accessable by car (if possible), to make it
easier for routing software. Of course that will direct you to an
arbitrary spot in a city with routing software, but it's just the same
with google if you want to go to "Paris, France" or wherever. Most
routing software will direct you to (or close to) the center of the town
then.

I would rather have that place to be the central square in a town than
some tiny cul-de-sac street which just happens to be close to the node I
entered as place=city;name=Paris.

Spaetz