okosm at johnfreed.com
Fri May 25 13:53:40 UTC 2018
I would generally agree with all your points.
A slightly more formal definition (though not fully rigorous) for me would
be: a circular route is one in which, from any boarding area, you can
return to the same boarding area without being forced to disembark.
I say boarding area rather than point because of the fairly frequent case
where the dropoff and pickup points serve the same area (such as a train
station) but are not necessarily identical.
The example I gave in the other thread, I believe, is marketed that way
because people are indeed supposed to leave the bus, though I would imagine
that since most people use weekly or monthly passes, most drivers would
probably look the other way.
Thorsten's description is good, but technically pretty much any bus route
is a roundtrip (circular) topologically. A typical route is:
A -> B -> C -> (forced disembark) -> C' -> B' -> A' -> (forced disembark)
-> A ...
Typically B and B' are on opposite sides of a two-way street. That's the
simplest example. Topologically it's the same as a circular route, because
the bus makes a U turn at each end and continues. So it's a question of
marketing and general public understanding, more than mathematical rigor.
But the variations are bewildering.
Personally, I track circular routes for validation purposes. As I
understand Public Transport version 2, every route variant requires a route
master, even if the master has only one variant. So masters with one
variant are oddballs; most routes have at least two. A master with only one
variant is typically either (a) a circular route (so I mark them in my
private database), (b) a PTv1 route (thus needing an upgrade), or (c) an
error (thus needing further investigation).
I would mark the OSM example above in my data as a circular route, so I
won't flag it as needing further work, but it would not meet my more formal
definition of a circular route, because of the forced disembark, so I would
not consider it as qualifying for a "roundtrip=yes" tag.
If there is sentiment to change the name of the tag, I would suggest
"route:circular=yes". (There are "only" 25000 in the OSM data, so it might
be manageable.) I don't like "circular=yes" because it's so vague.
(example: building=silo circular=yes ?)
On Fri, May 25, 2018 at 1:23 PM, Peter Elderson <pelderson at gmail.com> wrote:
> Oops, I didn't think this topic would generate so much response, even
> though I charged a bit in the first mail.
> Let me try to make some sense of it. I have seen enough use cases, I think.
> a. There are two use cases which use the actual definition on the wiki: a
> geagraphically closed route, start-point=end-point. One is about marking
> routes as roundtrips based on JOSM validation, then monitoring if the chain
> had broken so you can fix it. The other is marking an unfinished route as
> roundtrip in order to detect it for completion. To me, this is almost the
> same use case.
> b. A range of use cases are opposite: a geographical roundtrip has to be
> regarded as non-roundtrip, or a geographical non-roundtrip has to be
> regarded as a roundtrip anyway.
> Could we agree that the wiki should cover b.?
> I think this does not exclude a.
> If anyone judges that a geographical roundtrip should explicitly be tagged
> as roundtrip=yes, ok.
> do think that when one of the use cases under b. applies, then you have
> an exception to what the map says, with a reason. Then this takes
> precedence over the geographical default.
> This could be a geographical roundtrip tagged as roundtrip=no for whatever
> reason, or a geographical non-roundtrip tagged as roundtrip=yes, for
> whatever reason. It would be nice to know the reason, of course. For my
> part, "everybody/nobody here calls this a circular line" is reason enough.
> Could we agree on that too?
> If so, all that remains is add this to the wiki.
> Vr gr Peter Elderson
> Tagging mailing list
> Tagging at openstreetmap.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Tagging