[Talk-GB] Spurious No U turns in GB

[Robert Skedgell]

On 27/06/2018 16:01, David Woolley wrote:
> On 27/06/18 15:30, SK53 wrote:
>> # Junctions with islands which also cause a single carriageway to
>> split and merge around the junction.
>> #
>
> I would definitely agree that this is tagging for the renderer, and
> wrong.
>
> However, I can think of a case like this where you might well get
> attention from the police, for a dangerous manoeuvre, if you U-turned
> at any time but the early hours.  I seem to remember that one of the
> routers thinks this is a good place for a U-turn.
>
> Obviously the correct thing is for routers to use a heuristic to
> detect that if two one way roads join at a very acute angle, they
> shouldn't route from one to the other, or should give it a large
> penalty, but it does beg the question as to whether OSM should allow
> router hints.
>
> The problem with hints is that, like "too dangerous for cycles, so
> mark as no cycling" they are subjective, and the router should really
> be looking at the topology to infer a that this is really a dual
> carriageway.

Where there really is a no U-turn sign (and consequently an offence
contrary to s. 36 RTA 1988 for failure to comply), might it help to tag
it with traffic_sign=GB:614 (see Schedule 3 to TSRGD 2016,
<http://www.legislation.gov.uk/uksi/2016/362/schedule/3/made>)?

This wouldn't do anything to help routing algorithms, but might make it
a little easier for mappers to find potentially spurious restrictions to
re-survey.