[Tagging] noexit=yes on ways ?

[Florian Schäfer]

Am 09.04.2014 14:51, schrieb Ronnie Soak:
>
>
>     > There can be a way that IS connected on both ends and still is a
>     dead end. A
>     > road can end in a wall or a fence, where on the other side the road
>     > continues.
>     > There may be other tags there (barrier=*), but still it would be
>     hard to
>     > quickly spot the dead end side with noexit=yes tagged only on
>     the way
>     > instead of the node.
>
>     No. In such cases, only the "barrier" tag is important. No additional
>     tag required.
>
>
> A noexit=yes tag is still a good idea to communicate to the next 
> mapper that there really is no exit for any transportation mode.
> A second mapper may suspect a wall/fence/exotic barrier type/whatever 
> being still passable by bikes or pedestrians.
What about access=no instead of noexit=yes? This would be more accurate 
and can't be misunderstood so easily. As the discussion showed, some 
mappers used noexit=yes to tag deadends which allow pedestrians to pass 
at the end. access=no is clearer in this aspect and can cover several 
cases (only pedestrians can pass, only bicycles, ...).
> Also the barrier=* might still be missing, because the first mapper 
> only cared to map highways.
> same goes for the access:*=* tag. It might still be missing. Mapping 
> doesn't only come in nothing vs. perfect.
I don't understand your point. If the right tags (barrier and access) 
are missing, we should add noexit=yes?
What about cleaning up and add the accurate tags (barrier and access)?

noexit=yes is for situations where a way ends in the current data to 
communicate to the other mappers: "There is NO way of traveling further".
It is _not_ for places, where a highway continues in the data, but where 
there are access-restrictions!
> As a means to communicate an intention from one mapper to the next, it 
> simply is more clear when mapped on the node than on the way.
> I simply gave an example where the end of the dead-end way can not 
> simply be deduced by its geometry.
I'm totally with you on the first point. Nodes are much clearer for this 
tag, because the information, that a street ends for all transportation 
modes is a feature of an end of a street and _not_ a feature of the 
whole street (think of streets with multiple ends).
But I think in your example noexit=yes should not be used. As stated 
above, barrier=* and access=* are much clearer and fit the situation 
better. Additionally, noexit won't be recognized by a router.
noexit _only_ makes sense at endnodes of ways.

An example where the tagging of noexit on ways is not sufficient is a 
T-shaped deadend, where the crossbar is one OSM-way. At one end 
pedestrians can pass, at the other end not.

Regards,
Florian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20140409/c75c69d4/attachment.html>