[OSM-talk] [Tagging] - RFC - Motorway_link impliesoneway=??

[Matthias Julius]

Alex Mauer <hawke at hawkesnest.net> writes:

> On 10/3/2008 6:27 AM, Ben Laenen wrote:
>> Then read the example on
>> http://wiki.openstreetmap.org/index.php/Tag:highway=motorway_link once.
>> It says: "The green way in this example can then be a simple junction
>> withhighway=motorway_link without the oneway tag, as it is supposed to
>> be used in both directions."
>>
>> When you read that, you don't assume there's no implicated oneway value.
>> It says that by default it's oneway=no like with any other road.
>
> Fair enough, though as far as I know it's never stated anywhere else 
> that oneway=no is the default.  Certainly it's not on the Key:highway or 
> the Key:oneway page.

Well, a highway is either oneway or it is not.  And, I don't know if
this is stated somewhere, there is probably a general assumption that
any boolean tags default to no/false/0 if absent and no implied value
is explicitly specified.  Am I wrong here?

>
>> Note that the "implies oneway=yes" in the right column wasn't added on
>> the page until June this year. Only from that point onwards the page
>> started contradicting itself. Before that it clearly said oneway=no is
>> implied.
>
> Only since October 2007.  So there's a seven month window in which it 
> could have been taken to be oneway=no by default, though that was never 
> stated clearly.

No, motorway_link has existed before that wiki page was created and
never implied oneway=yes.

>
>> Hence the only option is to revisit all motorway links that don't have a
>> oneway value and add it. After that we may have a oneway=yes
>> implication.
>
> It's really not the only option.  We can also just accept that there are 
> a few motorway_link tags that are wrong.

We should never *accept* that there is wrong data in the database.

> Which there probably will be no matter what.

But, this is no reason to accept that.  Which means that we should
correct wrong data where we find it and not knowingly create wrong
data and just accept it as "collateral damage", IMHO.

> I'd much rather have sane defaults and a slight inaccuracy then have
> to deal with insane defaults forever just because of a short window
> created by someone who made some bad assumptions.

It might be a while before the fact that an implied value has changed
has trickled through to everybody.  Such a change should be announced
ahead of time after which all affected objects where the tag in
question is missing get it added with the old implied value.  Then the
implied value can be changed.

There will still be people who miss all that and keep on tagging as
before, but I don't see a reasonable way of avoiding that - except not
changing implied values.

Matthias