From OpenStreetMap Wiki
Jump to: navigation, search

This page contains outdated discussions, which were moved from Talk:Buses, discussions which predate Relations and other widely used conventions.

Storing Bus Routes with Bus Stops

Resolved: Use route relations

rjmunro proposes:

  • add another tag on bus stops called bus_routes that has a semicolon separated list of bus numbers (or names for routes without a generally known number like the Oxford Tube).
Can you give an example of the values for that tag? The problem I encountered when I started to do this is that you need a way to tell if the '300' bus route in St Albans is at all related to the '300' bus service in all the other towns/cities in the UK, let alone other countries. TomChance 13:02, 25 January 2007 (UTC)
  • If there are two bus stops opposite (or close to opposite), serving the same busses, they can just go as one node at the average point of the two stops.
  • If there are multiple bus stops serving different routes, one after the other (as in busy town centres), they can go in as one stop that just serves all the routes.
    • You can't say to do that (at least not for all cases), multiple bus stops can be round the corner or down little side roads. Coming out of a big London station to catch a bus it's great if you remember what letter stop to go to, finding it you really do need a map of the bus stops (thankfully a map poster will be up somewhere, but not yet on my OSM map).

Bus stops are geographical features which are pretty much the same from year to year and deserve a place in OSM. Bus services change very much more frequently and because of this I think should not be included in OSM's database. This is not to say that OSM should not be used for drawing the routes of bus services, but this would be done on a local (transient) copy. The routing information would be derived from an up-to-date external source, probably in a specialist format like TransXchange, which typically lists the stops served by the service. In order for this procedure to work properly, it is essential that the bus stops are nodes on the street, not off to one side. It is also desirable that they have a direction specified in terms of OSM's structures, as discussed below, rather than using an external convention. Chrismorl 15:48, 26 January 2007 (UTC)

The tagging of bus routes in bus stop nodes is obsolete. Please use relations of the type=route, route=bus.

--Lulu-Ann 09:56, 5 March 2009 (UTC)

Directions of Bus Stops

Disregard: Route relations and a node to the side of the road

rjmunro proposes:

  • If the bus stop is only for busses in one direction, it should have a bus_direction=(N|S|E|W), pointing in the (rough) direction that the busses travel. (not strictly necessary on one way streets, but still strongly suggested because it means a bus routing system need only check one tag). I've avoided segment direction because bus stops are nodes. Nodes don't have direction, and direction may get changed for other reasons without realising that there are bus stops.
What about directly NW, or NNW? :) Though it doesn't place them on the highway, to solve this, the one/both sides problem, and the people editing a way problem... I've been placing bus stops as isolated nodes either side of the highways so it's completely clear to me, until somebody proposes a better system. TomChance 13:02, 25 January 2007 (UTC)
I'd always pick the nearest direction of the 4, and use the below algorithm to convert to segment direction. One of the 4 directions will always be nearest. More than 4 directions complicates the below algorithm very significantly - you need to convert to mercator projection before you can do the comparison. Rjmunro 14:36, 25 January 2007 (UTC)
For a segment that joins the bus stop:

  if direction = "N": startnode = node with lowest lat
  if direction = "S": startnode = node with highest lat
  if direction = "E": startnode = node with lowest lon
  if direction = "W": startnode = node with highest lon
I would favour using the segment direction to give direction to a bus stop node. Since OSM already uses this system for oneway streets and rivers, it would be going with the flow, and wouldn't need extra conventions. It needs the node to be at the junction of two segments with the same direction, which would nearly always be the case and would be always possible. The concept would still work if even if segments were removed from OSM.
Each stop could have two tags: for_flow_direction = yes; for_antiflow_direction = yes. This example is for a stop which serves buses in both directions (as in happens in some rural areas); most stops would omit one or the other tag. The names of the tags are awful - can somebody suggest alternatives?
These tags would be also used for any other node types requiring direction (e.g. speed cameras) and editors like JOSM could provide a degree of mitigation of the corruption risk mentioned by rjmunro above by swapping the tags when the segment reverse command was used.
-- Chrismorl 15:48, 26 January 2007 (UTC)
There is a similar proposal at Proposed_features/Side_of_stop.
I'd be unhappy with making tags which relate to the direction of flow because it's all too easy for people using current editing tools to alter the direction of a way without being aware of what they might be doing to nodes along the way. Relating it to compass directions is awkward but robust in the presence of existing editors, and this is what I've been doing recently.
Another question for segment-direction proponents: if this Node is shared between Ways, to which Way does the direction relate?
--achadwick 17:23, 19 April 2008 (UTC)

Other things

A problem that we may have with features like bus stops is that they are also nodes within a way, and if an editing user is not being careful, they may be tempted to move them along the way to make a better curve elsewhere on the road. Can we add features to JOSM and the applet that warn you if you try to move a node in a way that has tags of it's own (other than created_by)? Rjmunro 12:33, 25 January 2007 (UTC)

Also, should we add something to routes that branch (e.g. _a, _b etc.) for the different branches so that we can tell which way each bus will go? Rjmunro 12:33, 25 January 2007 (UTC)

The bus stop near me has the identifier CH1393 on it. This is, I think, the "ATCO code". It appears in some Cheshire County Council data I have as 0600CH1393, and this becomes a 12 digit number in NaPTAN. This stop also provides CHWMADJ as an identifier to obtain timetable info by SMS. If a stop has an identifier like any of these it should be recorded in OSM, perhaps as ref = CH1393 etc, whatever the format, until we sort it out.
In the UK there is a national database of bus stop information, NaPTAN. It is of course hedged around with Crown Copyright, requires a licence and does not seem easily available. It is administered by Thales whose background in defence work, is not likely to help with access. The data is provided by local transport authorities (Cheshire County Council, where I live) and maybe they could be more easily persuaded to share this information. But the OS have stranglehold on the geographical information, so we cannot be too hopeful. Chrismorl 15:48, 26 January 2007 (UTC)

In Calgary each bus stop has a ref number, this is used both to identify the stop and to provide a 'call for information number'. It obviously make sense to record this in the ref tag, but how should it be handled when a stop has a cluster of signs with different numbers (example)? --Mungewell 03:56, 18 September 2008 (UTC)

(UK) Bus stops don't appear in the key to the public map page UrbanRambler