[OSM-talk-be] busroutes

[Ben Laenen]

Maarten Deen wrote:
> Have a look at http://wiki.openstreetmap.org/wiki/VRS , for a lot of
> relations, there are two routes. Often also tagged with a from and to in
> the relation, although I don't know if that really helps in a program.

"a lot of relations": I count 12 on that page with two relations, given the 
number of routes on it I wouldn't really call that "a lot" :-)


Anyway, I personally don't see why splitting the relation in two would be so 
much better. Basically the only reason why you'd do it because it tags a 
little bit easier since you don't have to add 'forward' and 'backward' roles. 
It doesn't solve anything else: the ambiguous possibilities to follow a route 
when a bus makes loops are all still there for example.

IMHO, the best option would be to stick with just one relation, but map it 
differently: take a starting point on the bus route and then just add the ways 
in the order the bus follows them to get to the other end, and then add the 
ways in order as the bus goes back to the start point (so usually adding the 
same ways to this relation again). In principle, you don't need 
forward/backward tags with that either.

Alas, we have problems with this since one of the main editors can't handle 
ways that belong multiple times to the same (so if someone else e.g. splits up 
a way in that editor the relation is broken), and it doesn't keep the order of 
the members in the relation.


I have the impression that making two relations of them is trying to patch 
this: avoid ways that belong multiple times to the same relation by putting 
them in two different ones. This doesn't work properly btw, since I know bus 
routes that go up and down the same road in both directions between the 
termini. One could suggest the topology of the ways belonging the unordered 
relation would always make it possible to get the exact way order the bus 
follows, but I'm not really convinced of that yet. Certainly if you drop the 
forward/backward roles: then you really cannot know anymore in which direction 
the bus rides a loop in its route. And for now I can see only one thing two 
relations without forward/backward can represent more that one with 
forward/backward: if the bus follows a loop in one direction, but doesn't 
follow it in the other. But if that's worth it to start tagging something 
differently? Better to wait for when Potlatch can finally handle relations 
nicely and map the bus routes properly instead of going to this intermediate 
method that doesn't bring much advantages and brings its own problems.

Greetings
Ben