[OSM-talk] How to use lanes= for two way single track roads?

[Andy Robinson (blackadder)]

Jo [mailto:ml at winfix.it] wrote:
>Sent: 13 March 2008 10:13 AM
>To: Andy Robinson (blackadder)
>Cc: 'Bruce Cowan'; 'DavidD'; talk at openstreetmap.org
>Subject: Re: [OSM-talk] How to use lanes= for two way single track roads?
>
>Andy Robinson (blackadder) wrote:
>> Bruce Cowan wrote:
>>
>>> Sent: 13 March 2008 2:28 AM
>>> To: DavidD
>>> Cc: talk at openstreetmap.org
>>> Subject: Re: [OSM-talk] How to use lanes= for two way single track
>roads?
>>>
>>> On Wed, 2008-03-12 at 21:09 +0000, DavidD wrote:
>>>
>>>> According to map features the value for the lanes key should be.
>>>>
>>>> "Number of travel lanes in each (or only permitted) direction"
>>>>
>>>> I've been tagging to this definition.
>>>>
>>>> "Number of travel lanes on the way"
>>>>
>>>> This makes more sense to me because you can tag two way single lane
>>>> roads with lanes = 1. The map features definition on the other hand
>>>> doesn't give an obvious way to tag these roads. lanes = 0.5 perhaps.
>>>>
>>>> I didn't pick up the lanes tag from map features but some other wiki
>>>> page. I'm sure the wiki page defined the lanes key similarly to my
>>>> definition. Unfortunately I know can't find that wiki page to check.
>>>>
>>>> Was the lanes tag added to map features wrongly?
>>>>
>>> While thinking about the many roads up here that are single tracked,
>>> it'd be quite useful to be able to map them properly. The result was
>>> passing places [0]. I am under the impression that lanes=1 means 1 lane
>>> each way, but this makes these roads impossible. I'd quite like another
>>> tag, such as narrow=true (or something non-boolean) which would mean the
>>> lanes number is number of lanes overall as opposed to in either
>>> direction.
>>>
>>
>> This why the original intention was that lanes was equal to the number of
>> running lanes on the physical way. In your case it would be lanes=1
>except
>> (if you like) at passing places where lanes=2. Since a passing place is a
>> very short length normally it would be simpler to tag a node with lanes=2
>at
>> each passing place. Isn't that the simplest method and easily
>understandable
>> in the data?
>>
>That seems a lot more logical than what is done now. lanes is the
>physical number of lanes on a road/carriageway. What would be nice would
>be the ability to also indicate the direction and also whether it's a
>lane for making a turn left or right or going straight ahead. It should
>of course be possible to have more than one lane for turning/going
>straight as well.

It's not at all easy to tag usage restrictions on ways or even nodes by
themselves and was one of the reasons that relations was set up.

I don't believe we have yet come up with a simple way of defining turn
restrictions, we had lots of discussion about it and there is a perfectly
good way of doing it with relations, but none of it is "easy" for the less
experienced mapper, yet routing information is still one of the most talked
about aspects of the project and the one that will hopefully eventually put
OSM on the same playing field (in terms of capability) with the main
opposition.

Personally though I think its time will come. SteveC and I were discussing
this week what we do when we are "mapped out", ie when it is no longer
economically viable (time mainly) to map anything in our local area because
its all been done. That's fast approaching for those of us who have been
mapping a fair while now. The next logical step is to rethink the need and
refocus on making the areas already mapped better, and that will include all
the turn restriction data.

For now though I think our focus is quite rightly getting a more complete
coverage of basic data, although I'll warmly support anyone who is grabbing
their own area by the scruff of the neck and getting better routing data
into it.

Cheers

Andy

>
>Jo