[Tagging] Dispute prevention: meaning of lanes tag

[Andrew Errington]

On Fri, 27 Apr 2012 18:54:26 Ilpo Järvinen wrote:
> On Fri, 27 Apr 2012, Andrew Errington wrote:
> > A lane count of 1.5 is very confusing.  What does it mean?  What is the
> > width of each lane?  Is it really 1.5?  Should it be 1.55, or 1.4, or
> > 1.6?
>
> ...No, it's not multiple of some magical "default lane width" like you
> imply. But simply _something_ between "normal" lanes=1 and "normal"
> lanes=2.

But the width=* tag tells you this.  Specifically, the width tag tells you if 
there will be a problem *for you*.  Since I have never met you and I don't 
know what vehicle you are using it would be presumptious of me to tell you 
that there are 1.5 lanes.  Also, it doesn't make sense to allow lanes=1.5 but 
deny 1.55, 1.4, 1.68 or any other fractional value.  What you are doing is 
introducing a 'special' value with a special meaning.  I think we should try 
to avoid having to interpret special cases.

> > It's more useful to tell me width of the road.  Then, if there is one
> > lane I can see maybe it's very wide, or if two lanes I can see maybe
> > they are very narrow.
>
> ...But how can I tag you this: A road which is lanes=1+wide _AND_
> lanes=2+narrow at the same time? ...You ask me to provide width and select
> one of those two, and that is what I oppose, unless you give me some real
> tag that is not width to tell that 'hey, there really isn't lanes marked
> (which makes it kind of lanes=1) but two can somewhat fit (which makes
> it kind of lanes=2 but not really because it's only "somewhat")'!
>
> ...What I would not want to do is to tag those lanes=1 because that's
> certainly a lie as anyone can clearly see after observing some
> bidirectional traffic there.

It's not a lie.  A single lane may be bidirectional.  In fact, in this case 
you *should* tag it lanes=1.  If oneway=yes is not present then it means one 
bidirectional lane.

> > In summary, I think simpler is better.  A non-integer lane count is
> > useless. Use the width tag.
>
> I oppose using width tag (at least alone) for this because it won't
> convey the double meaning. Some other tag than width and tagging with
> lanes=2 perhaps (like I already suggested much earlier)?

I don't think there is a double meaning.

lanes=1 tells me there is one lane.  It does not mean one direction, nor 
should anyone assume that.  I *think* you are saying that lanes=1.5 tells 
me "this road is not really wide enough for two-way traffic, but there *is* 
two-way traffic so if there is a car coming the other way you have to 
wait"[1].

For the purpose of discussion, let us assume that a road of 2.5 m width is too 
narrow for cars to pass.

lanes=1 width=2.5 tells me the same thing (one narrow lane, cars travel both 
directions, but only one direction at a time).

lanes=2 width=2.5 tells me the same thing (two very narrow lanes, cars travel 
both directions, but only one direction at a time).

lanes=1 tells me the same thing (one lane implies cars cannot travel both 
directions at the same time, but no oneway=yes tag implies cars can travel in 
both directions.  We don't know the width but there must have been a reason 
for a mapper to tag it with lanes=1).

width=2.5 tells me the same thing (no lanes=* tag and no oneway=yes tag 
implies two lanes, but both must be very narrow therefore cars can only 
travel one direction at a time).

I don't think any of these assumptions are unreasonable, and they don't alter 
the existing meaning of the tags, which I believe are already quite clear, so 
we should use them and not alter them with special cases.

Best wishes,

Andrew

[1] I have made this sentence to interpret what lanes=1.5 means.  If my 
understanding is incorrect please state what lanes=1.5 actually means.