On Thu, 23 Feb 2023 18:39:19 -0500 Greg Troxel <g...@lexort.com> wrote:
>Niels Elgaard Larsen <elga...@agol.dk> writes: > >> We have to accept that the tagging is never complete. And when >> surveying, it is often easier to tag "locked" than "access" (we can >> se the lock or try to open the gate but there are often no signs). >> So the tagging might reflect that we know that a gate is usually >> locked, but we do not know who can use the gate. >> >> At least "locked" should imply access=destination or private for >> routers. > >I don't see it that why. access=private, probably. access=destination >means you can use a way if you decide to go someplace that you need to >use the way to get to. But that's wrapped up with can you. If a way is access=destination, a router should only take you there if you have set your routing destination close by. And if it is your destination, most likely you have the necessary keys or someone will let you in, because you live there, you are a customer, you are visiting someone, you deliver something, you intend to pay for access, etc. >It's a little unclear to me what a "locked=no" gate is. Here is one that I use daily: https://www.openstreetmap.org/node/3498950779 It has a lock, but is never locked. The fence is only a meter high, and you need to pass the gates to access the main entraces of Koldinggade 20-24. I do not know why these two gates have a lock. Probably the company setting them up just used the same handles and locks as the two higher gates just south of there. Those two gates are locked. > I'm guessing >it's a physical barrier than anyone can easily move out of the way. So >arguably a barrier without locked shouldn't preclude routing, just a >2-minute delay. > >Then there is 'access=' on the barrier. This doesn't make a lot of >sense to me as in practice access is about the way, and the gate is >either: > > not locked, and not an impediment, but might keep cows in > > locked, and an enforcer of something that is alraedy true > >or > > locked, and enforcing no travel when people have a legal right to do > so. Still, routers should not attempt to use this. > > >So maybe: > > foot=yes on a barrier=gate means that a person on foot can pass the > gate (perhaps by walking around it) with minimal difficulty, such > that the gate's presence should not affect routing > > barrier=gate locked=no (or no locked, default?) means that all modes > may physically pass, with a mode-specific typical cost > > barrier=gate locked=yes means that all modes may not pass, unless > there is a mode-specific foot=yes or bicycle=yes > > >The real problem is that locked refers to the ability to open the gate, >but many modes pass without opening, and we really care about "can this >mode actually traverse". so locked=yes is a shorthand for all modes >of travel and we don't have a foot:can-walk-around=yes tag. Until we >deal with that head on, this is going to be messy. > >> According to the wiki, motor_vehicle=no means that motor vehicles are >> not allowed to travel through the barrier. The wiki does not say that >> having a key to the lock changes that. > >agreed. > > >So we need three properties: > > legal right of access, perhaps only needed on ways > > physical ability to pass a gate > > is the gate locked, and if so which modes does that apply to > _______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
