On Tue, 7 Jan 2020 at 18:05, Volker Schmidt <vosc...@gmail.com> wrote:

> I don't see misplacing nodes as being a good alternative to getting the
>> routeing right.
>>
>
> At least in the context of the legal requirements in Italy, I am not
> suggesting to misplace a node. I suggest to put it where the house number
> has to be (in Italy), i.e. on the entrance from the public road.
>

I think that interpretation conflates two separate issues: the address of a
building
and the placement of the house number.  We do not conflate road junctions
with
nodes marking the signs for road junctions (which some people do not map
anyway).  It is the building that has an address, not the sign.  The sign
is an
indication of the address.  The fact that Italy demands a specific
placement of the
sign doesn't really matter because if you know where the building is and
have roads
(and possibly drives/footpaths) mapped then you can figure out where the
sign is
going to be (but for multiple properties behind a single gate the reverse
is not
true).

>
> Regarding adding the (private) footpath or driveway from the gate to the
> house, I did not expect a routing algorithm to be so intelligent that when
> routing by car it also takes into consideration additional bits at the end
> that are private and/or pedestrian. But I am not a routing expert, may the
> AI has done wonders there as well.
>

Google certainly manages it, and can give you a route which involves buses
and
walking.  In any case, in urban areas (at least in the UK) residential
streets usually
have names: if a house address is 7 Foo Street then it's almost certain
that it is
reached by going to Foo Street, even if the placement of the house on the
lot puts
it nearer to Bar Street.  I would definitely expect a routeing algorithm to
take account
of street names, where present, even if it ignores driveways/footpaths (but
why
would it, since they're still ways?).

Yes, there are some weird exceptions near me.  Rare exceptions.  The UK
seems to conserve postal addresses even if road widening and re-routeing
means those addresses are misleading.

But why do we need to have the full street address on the building at all?
>

To identify it.  In the UK, house number or name, plus postcode is
sufficient to
uniquely identify it.  People, however, still find other information
useful.  Such as
the address being 7 Foo Street means it's probably accessible by Foo Street.

The rest is more for completeness and acting as a reverse gazetteer; that's
the
building, what's the full address?  It's useful, so I add street name and
town even
though house name/number and postcode is all that's strictly necessary.

Oh, and it may also help disambiguate nominatim queries if somebody knows
house
number, street name and town but not the postcode.

Nervertheless I admit that there will certainly be cases where we need some
> way of tying together the point where the navigation device finds the
> address and the buidling where the people live whom you have come to visit
> to have a cup of tea. A site relation ?
>

I would say that mapping driveways/footpaths is adequate for both humans and
routeing algorithms.  In many cases we don't even need that much: named
streets
give all the information needed.  Maybe, in exceptional cases, but as a
default
thing for every building in Italy it seems excessive (especially as some
mappers are
scared by relations).

-- 
Paul
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging

Reply via email to