[Talk-GB] traffic island mapping / harmful detail?

[Colin Smale]

>     On 03/31/2021 3:12 PM Edward Catmur <ecatmur at googlemail.com> wrote:
> 
>     On Wed, 31 Mar 2021 at 13:25, Colin Smale <colin.smale at xs4all.nl mailto:colin.smale at xs4all.nl > wrote:
> 
>         > > This is fine for straight-across crossings with a simple island. What about a longer island, with staggered crossings, and independent traffic lights for the two halves? It may be only elongated by a couple of metres, but topologically speaking (relevant for routers/navigators, and don't forget the pedestrian's viewpoint is different to the driver's) very different. Maybe in this case it would be correct to split the way around the island. If so, what makes the difference? The stagger? The lights?
> > 
> >     > 
>     I'm not sure whether it's strictly valid, but one possibility would be to have a (Z-shaped) crossing way that overlaps with a short section of road. For the lights you could have 2 nodes, with direction=forwards/backwards.
> 
If you submit the roadways on each side of the island to duck-testing, they look just like carriageways... Intended for motor vehicles, kerbs on both sides, "keep left" at the split... Your suggestion is just a work-around to get some of the detail in without addressing the elephant in the room.

What we are discussing here is the level of detail. It cannot be incorrect to split the way into two short one-ways, as that is a matter of objective fact, because that is what we have. Whether it is worth the bother, is subjective however. If someone suggests a rule that can be applied objectively, such as "splits followed by joins within ±10m don't count" then we can discuss that.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20210331/321b9b6a/attachment.htm>