[OSM-talk-be] busroutes

Ben Laenen benlaenen at gmail.com
Thu Aug 12 09:34:30 UTC 2010


On Thu, Aug 12, 2010 at 10:48 AM, Ivo De Broeck <ivo.debroeck at gmail.com> wrote:
> I do not understand this. I have used Potlatch and have put in disorder all
> members of the relation. Then i have used the relationchecker and all
> members came in the right way. This was for a oneway buss-route.

The relation analyzer will reorder them on the page, but they are not
in order in the database. There's btw an equally great chance that the
analyzer will return them in opposite order from end to start (it
depends on which way it chooses first and which it selects after that,
without changing the relation the analyzer will always give the same
order normally). The order you get on that page is not how they're in
the database. You can check the real order with JOSM or just in the
browse pages of the API.

> When i
> tried to put the backward route, all members were in disorder.
> This is a normal buss-route: (F=forward / B=backward)
>                       B
> begin --F/B-----<>------F/B----end
>                       F

You have to know what the analyzer does and doesn't do. So in short:
it groups the members with the same role (so, all "backward" will be
grouped separately from all members with an empty role, etc), and then
it reorders the ways so each group gets as large routes as possible
(not exactly like that, but it'll do as explanation).

So since we have forward/backward/empty roles, it will automatically
not give good results. You'd have the same problem if you'd add only
one direction of the route and added backward/forward roles (try the
analyzer on some local walking routes around Antwerp for example).

Basically, the analyzer isn't really good for two-directional routes.

> Can we discuss further
> at http://wiki.openstreetmap.org/wiki/Talk:WikiProject_Belgium/Conventions/Bus_and_tram_lines
> ??

I don't see a reason why we'd move discussion over there? It's more
visible over here and more people are likely to participate in the
discussion here.

Greetings
Ben




More information about the Talk-be mailing list