[Talk-GB] Cycle lane in one direction only

[Andrew Chadwick (mailing lists)]

Ed Loach wrote:
> What is it we are trying to address here exactly? I'm assuming it is
> cyclelanes that are part of the road/way as

This is with reference to

   http://wiki.openstreetmap.org/index.php/Key:cycleway

It is currently *impossible* to represent a way which has a cycleway 
down just one side of it on a road which is not oneway=yes|true|-1. That 
is what Leedstracker was describing; I think the currently documented 
practice misses an important case for the UK.

> if they are separate
> then they should be added as a separate [highway=]cycleway (which may or may
> not be oneway).

Yes, specifically just the ones that are on-road or on an adjacent track.

> Assuming that we are trying to tag cyclelanes marked as lanes on
> roads, then we need to consider whether we can safely assume that
> cycle lanes drawn on the left of a road (in UK) will always be in
> the same direction as the traffic flow.

We cannot. Google "contraflow cycle lanes", and consider roads like 
Little Clarendon Street in Oxford (one-way street with a cycle lane in 
the opposite direction, on the right from a driver's POV).

> If this isn't a safe assumption, then we might need to consider a
> scheme which indicates both which side of the way a cyclelane is,
> and which direction cyclists will be travelling relative to the
> direction of the way (with/against or both).

Is a goat of a job. But that's the jist of it.

Side of the way relative to traffic is less important than side of the 
way relative to the direction of the way's "arrow", certainly from the 
rendering perspective if you're putting nice casings one one side of the 
way, the other or both.

It's unhelpful from the routing perspective, of course. However, the 
side of the road one cycles or drives on in a given country is better 
left to the routing logic, IMO.

> So if we assume we have a directional way (as per oneway=yes, but in
> this case not a oneway way), then we could have cycleway:left to
> indicate that the cycle lane(s) are only on the left of the road.
> But we may need to specify that the cycleway direction is opposite
> to the traffic flow, same as traffic flow or two way.

I suppose that for my example above, I might do

   highway=unclassified
     ;; it's a small road
   oneway=yes
     ;; ... which is oneway for all traffic
   cycleway:$RIGHT_WRT_WAY_DIRECTION=opposite_lane
     ;; except for bikes on an on-road painted cycle lane going
     ;; in the opposite way to the direction specified by
     ;; the oneway tag

which actually completely specifies everything. For oneway roads, you 
have everything needed for routing or rendering right there, no 
inference or external knowledge needed.

The same isn't quite true for bidirectional, non-oneway roads with a 
bike lane on just one side WRT to the way arrow. Everything's there for 
rendering, but for routing bicycles routing software would also need to 
know which side of the road people cycle/drive on, assuming that the 
cycleway is oneway.

Fancy logic like that is best left to routing software which already has 
a lot of it and the capacity for adding more of it easily than to 
rendering software which probably doesn't.

> Two examples. Botley Road in Oxford has cyclelanes (part on road and
> part on pavement) on each side of the road in the same direction as
> travel as the cars on the road. Centenary Way in Clacton though has
> what I will (when I get around to tracking it) draw as a separate
> cycleway as there is some grass verge between it and the road (and
> shared pedestrian use), but this is only on one side of the road and
> is two way. If there were no verge I'd probably want to try and use
> some sort of suitable tagging of the way.
> 
> Perhaps not the best of examples, but I'm trying to show that we may
> need to indicate which side of the road cyclelanes are as well as
> which direction you can travel on them.

Basically, I agree. I also think it's way more flexible if we represent 
which side(s) of the road they're on with respect to the way arrow.

Currently, which-way-you-may-cycle on a cycleway is represented by it 
being an opposite_foo or just a plain foo. I propose we address your 
double need by introducing new tags to say which side of the road it's 
on: a fact currently left unstated.

-- 
Andrew Chadwick