[Tagging] new access value

Fri Oct 9 21:26:22 UTC 2015

On 08.10.2015 12:16, Lauri Kytömaa wrote:
(...a lot. I try to narrow it down to the critical passages.)

> sometimes the signs restrict the
> traffic by "who", not by destination. I.e. in the Anreinerverkehr case,
> [...] it's no longer about the destination, but which
> group of people you, or the passengers, belong to. It's not about the "for
> a purpose (like agricultural, forestry, delivery), nor "by permission", and
> it's not about how your vehicle is registered as (the key part)
> 
> The number of different access tag values should be kept to a minimum,
[...]
> There's already a list of "by use" keys in on the Key:access page, although
> only "disabled" is directly comparable;

The access tags are built like this:
<mode of transport>=<who, or under which circumstances>

"disabled" is not a mode of transport. Therefore, it should rather be a
value than a key. The reason it is a key is because it was introduced in
OSM's stone age.

Let's have a look at the currently documented values:
yes, private, no, permissive, agricultural, use_sidepath, delivery,
designated, dismount, discouraged, forestry, destination, customers.

Some belong to the "purpose" category, but others belong to the "group of
people" category, and some values don't fit in either category.

yes ... (matches any category)
no ... (matches any category)
private ... owners and people with granted permission => group of people
permissive ... all except people with revoked permisson => group of people
agricultural ... purpose
use_sidepath ... same as "no" + hint to traffic sign
delivery ... purpose
designated ... same as "yes" + hint to traffic sign
dismount ... same as "no" + hint to traffic sign
discouraged ... same as "yes" + hint to traffic sign
forestry ... purpose
destination ... route geometry
customers ... group of people

So we've got 3x "group of people". The new tag will just be the forth in
that category.

> - however, if there are reasons why setting a destination there is subject
> to "doesn't belong to a group of people" limits, let's invent a new key,
> or a group of keys.

If you like to move the "group of people" category to new keys, you need to
deprecate access=private/permissive/customers to stay consistent.

> Those tags could then be used to tell that if the user
> has a destination there, they can go, but there may be reasons that they
> legally can't set a (motor) destination there.
> 
> So, the tags, for example:
> 
> motor_vehicle=destination
> + destination:limited=Anreinerverkehr

Non-English values with uppercase letters? Are you serious?
You misspelled it, and other will misspell it too. And nobody except
German-speaking people will have an idea what this tag means. Even among
them, there's a lot of misconception and dispute regarding that word.

> (latter would apply to whatever mode has =destination)
> + destination:limited:something=Johannesbach fishing resort
> when not obvious and when necessary to describe which is the only
> possible point of "contact" the local system requires.

How are routers supposed to make use of that free-text tag?
What does the "something" part mean?
I am quite puzzled, and I fear that data consumers will be too.

> or, if necessary (doubt it, but for future reference)
> destination:limited:motor_vehicle=Anleiter
> destination:limited:hgv=(something else)
> 
> The values can be indexed and explained in the wiki, and shown to the
> user as-is or with explanations.

Shown to the user? At what stage? And remember that applications are not
limited to route calculation. Displaying lists with explanations for each
single road may get difficult in a paper map.

> The second possible solution is to use only tags that define the restrictions
> group by group:
> - motor_vehicle=private
> - limited:Anreiner=yes
> - limited:Anreinerverkehr=no

Non-English words with uppercase letters even in keys?

> *
> - motor_vehicle=destination
> - limited:Anreiner=yes
> - limited:Anreinerverkehr=yes
> 
> Here, the prefix "limited:" is used to tell that the latter part is a name of
> a group. limited:disabled= and limited:customers= would be possible
> (this is why I opposed using customers as an access value, it's not a
> legal access type but a group).

Anrainerverkehr is not a group. So it does not fit in your tagging scheme.

"limited" sounds like a shortage. When you suggest a tagging scheme about
groups of people, you better name it like motor_vehicle:for=* (compare to
social_facility:for=*).
But it won't work either way, because you are soon getting 2-dimensional.
E.g. for group A it may be vehicle=yes and for group B it may be
vehicle=destination and for group C it may be vehicle=no. You need to
incorporate your scheme into
http://wiki.openstreetmap.org/wiki/Conditional_restrictions.

Anyway, it's already too late. The ship has sailed. You opposed "customers",
but you did not stop it. Let alone "private", which stands for a group as
well. You will never get access=private deprecated.

> For what's it worth, there's a remotely similar case here in Finland: we
> have lots and lots of no motorized vehicles signs (sometimes together
> with a more relaxed "driving to premises allowed", which on the other hand
> indirectly requires that the destination is beside the road on private property,
> if driving there ever were to be a subject of a legal discussion), with an
> exception literally translated as "maintenance traffic allowed") which has a
> definition that includes all sorts of uses beyond actual "maintenance", like
> more than one under seven year old children for each adult in the car, and
> delivery traffic, and transporting any mobility impaired persons, and picking
> up or dropping of taxi passengers (but no taxis otherwise), and other cases.
> It's not really possible to use existing tags accurately for all these, so some
> of us have just used the tag Fi:huoltoajo=yes; without or with
> motor_vehicle=destination if there's a more relaxed allowance, too. These
> are, however, mostly short service roads and ways where nobody would end
> up by mistake. Marking all of those as "destination" would be wrong - not
> everybody may drive there - and using 6+ tags would be pointless for
> mapping when they weren't even supported.

Chicken-and-egg problem. In order to get them supported, you need to propose
them first of all. When the tags get in use or get documented in Map
Features, support will become more and more likely, as opposed to
Fi:huoltoajo, which will never be supported by any international
application, because nobody knows that this tag exists, and nobody knows
what it means. As I already replied to Richard Fairhurst, don't use
proprietary tags. OSM is an international project, and everyone should use
tags that others can understand.

6+ tags sounds a lot; but looking at your examples, we already have a tag
for delivery traffic (access=delivery) and for mobility impaired persons
(disabled=*), and we may well make up tags for the other things too if we
try. When we can define a meaning, we can define a tag.

-- 
Friedrich K. Volkmann       http://www.volki.at/
Adr.: Davidgasse 76-80/14/10, 1100 Wien, Austria