The first example wasn't broken, IMO. It just shows that there are ASLs on several approaches to the same set of lights. I've fixed all of the other examples except the last one. These were mainly situations where the constituent highway had since been split and there isn't editor support for knowing only to keep the way attached to the road, thus it's easy for there to end up being more way members of the relation that there should be.
I agree that you cannot determine the ASL position itself with this scheme, but that's easily rectified by using some roles. E.g. include mark the role of the junction's node as 'junction' and add a further node at the position of the ASL and mark this 'asl', ensuring that this node is a member of the way approaching the junction. If we need to show that an ASL is just for certain classes of traffic then I'm sure we could just use the normal access tags, i.e. bicycle=yes, psv=yes, etc. The only kind of example that I can think of is in an experimental traffic scheme in Canterbury at the moment where buses / taxis have an advanced waiting position for going through some traffic lights. There's also a restriction to buses / taxis for going through the lights in just that direction, so it might be argued that it doesn't add much value by recording the "bus / taxi" ASL. I'll try to remember to snap a photo of the arrangement at some point (it's post Google StreetView's visit, having only started on Mar 27th this year). I've seen some junctions where the ASL's actual position is actually quite a long way from the highway=traffic_signals node in the past, and the use of the relation does seem useful to preserve the semantics of the ASL's relationship with the junction, where the "nearest junction" interpretation on cycleway=asl might prove ambiguous or even misleading. I'll try to dig out some examples. Gregory -----Original Message----- From: Andrew Chadwick (lists) [mailto:a.t.chadwick+li...@gmail.com] Sent: 23 May 2012 12:02 To: tagging@openstreetmap.org Subject: Re: [Tagging] Merge cycleway=asl/advanced_stop_line/bike_box/...? On 23/05/12 10:52, Gregory Williams wrote: > There's also > http://taginfo.openstreetmap.org/tags/type=advanced_stop_line > which uses a relation containing the way approaching the ASL and the > node of the junction of the ASL. With this method it's easier to > determine the direction of approach to the ASL than just tagging a single node alone. Thanks for the reminder; we'll have to organise those too. I'm not sure they're completely consistent: http://www.openstreetmap.org/browse/relation/75707 - 3 ways, all connecting to a central node http://www.openstreetmap.org/browse/relation/87140 - 2 ways, 1 node; ways are in sequence, node is *not* part of one http://www.openstreetmap.org/browse/relation/106889 - two 2-way sequences, 1 node, 1 plain way: 6 members http://www.openstreetmap.org/browse/relation/106868 - 7 ways all in a line, 1 node http://www.openstreetmap.org/browse/relation/158262 - 1 way http://www.openstreetmap.org/browse/relation/185408 - 1 node http://www.openstreetmap.org/browse/relation/239790 I assume the model was for 1 node and 1 or more ways connecting to it, and that the weird ones are degenerate cases. It seems rather subject to editor rot: though that's the problem with all relations of course. You cannot determine the location of the ASL itself with your scheme, which is a bit of an oversight for a geographic database. Additionally it does not say what sort of traffic the advanced stop line is for. I think for it to be fully useful it would need to state both. Still, I'm not averse to relations provided they build on simpler elements which are easier for newbies to add, and are *only* deployed to resolve cases of ambiguity. I think there's likely to be very little ambiguity in practice; my gut feeling is that simple tagging of locations would be capable of expressing the vast majority of cases unambiguously. Counterexamples welcome, of course! At this early stage we get to stipulate that there either MUST[1] or SHOULD[1] be a highway=stop line "nearby, as part of the same way", and that both nodes must occur on one, and only one way. Would that be a workable approach? Do situations like -------stop-----[hugegap]-------asl---junc---stop--------- ever occur in practice? I note that even this ambiguity can be resolved by breaking the way at the junction node. [1] ... see RFC 2119 ... -- Andrew Chadwick _______________________________________________ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging _______________________________________________ Tagging mailing list Tagging@openstreetmap.org http://lists.openstreetmap.org/listinfo/tagging
