[Tagging] no_u_turn restrictions for every entry/exit into a roundabout when the way is split because of physical separation?

[Kevin Kenny]

On Thu, Apr 5, 2018 at 6:06 PM,  <osm.tagging at thorsten.engler.id.au> wrote:
> I tought it was obvious, but let me spell it out: such restrictions
> represent a default which we should be recorded somewhere (not necessary
> inside OSM) once and observed by data consumers, not by creating potentially
> 100000’s of relations to again and again encode the default behaviour.

Uhm, of course.

That said, is it worse to have redundant data, or no data? The best
outcome would be that someone would finally have the discussion of
traffic restrictions on an area. (That's what would be needed to model
a lot of communities near me, which often have signs on all the roads
coming in, with text like: "Welcome to Niskayuna. Town speed limit 30
mph except as posted. No parking midnight-6am on any street after 4
inch snowfall." This is surely field-observable, and it applies to an
area (or to the ways bounded by that area). Given the field
observability, I think it belongs in OSM if mappers want to map it and
routers want to consume it.

Until we've had that discussion, and arrived at a consensus on what to
do about locality-wide signed traffic restrictions, we're offered the
choice of redundantly encoded data or no data.

I don't know what the right answer is, but it surely is not simply to
tell mappers "don't map it that way" without offering an alternative.