Which is better to represent a turning lane:
Obviously the second, because it is live, not just proposed, and provides more information.
Incidentally, on a quick look, I can’t see why the two are alternatives; it seems to me that if you are interested in micro-mapping road geometry, using both together would be better.
The first this is a simple, unstructured, junction layout, superimposed with a proposed feature that plots the physical boundaries of roads as connected areas, with an area:highway key to indicate that it should be interpreted as an alternative representation of the highway.
The second this is about encoding turn lanes as attributes of their parent road, until they actually break away from the road, rather than mapping them as distinct roads earlier on.
Please not: You can’t compare both wiki entries!
The second one adds information about lanes to the highway lines, so that routers and navigation apps can use them.
The first one needs still the highway lines. It just extends the lines with area mapping (no lanes mapping!) which is currently only used by some renderers as far as I know. Also the area:highway has no further tags and doesn’t copy the tags from the lines.
I just used the first as an example (please consider only the image directly linked and not the rest of the article). I’m only wondering about the representation of turn lanes, as a divided line that begins at the start of the turn lane, or as tags representing the turn lane followed by a line that starts at the point a physical barrier actually divides the two roads.
I’m not sure how I could do so, they seem incompatible to me.
In the first example, the line of the turning lane splits from the main way at the beginning of the turn lane, leaving no room to apply the turn:lanes tag anywhere.
In the second example, one can assume that the exit lane would be mapped as a line starting at the point a physical barrier divides its path from the main road, and that the entire stretch of the main way with a turn lane would receive a turn:lanes tag (and also a different value for the lanes=* tag).
I think the question is important because I’m seeing people using both, and where I map I’ve seen some people undo one form to do the other without explanation - so, edit wars ensue.
As the others said, only split way at the moment there is a physical divider. A continuous white line does not count as physical divider. Use change:lanes to indicate that you cannot switch lanes.
All routing apps I know of assume that the ramp starts at the physical divider. This is similar to the behaviour of non-OSM based apps.
If I see people that use ways to represent lanes, I contact them and tell them not to do it. Since the majority of roads are mapped as 2 in my part of the world (and in the neighbouring countries), it’s easy to convince them.