Missing/duplicated branches in /Line/Route/Sequence

Unified API
1

There seem to be a number of problems in this area, starting, I believe, from the addition of national-rail services to the API.

An initial simple example is bus route 358,inbound. There are currently 3 branches: 0, 1 and 3.

 0: Crystal Palace Bus Station-Sevenoaks Road/Orpington War Memorial, prevBranches:none  nextBranches:1,2
 1: Sevenoaks Road/Orpington War Memorial-Tubbenden Lane, prevBranches:0  nextBranches: 3
 3: Tubbenden Lane-Orpington Station, prevBranches:1, nextBranches:none

Branch 1 runs via Walnuts Centre. I am assuming that the missing branch 2 is the direct route, not via Walnuts Centre, to Tubbenden Lane and that branch 3 should also have prevBranches 1,2

This problem is repeated in routes R5, R8 and R10.

Turning to national-rail and using greater-anglia, inbound as a (relatively simple) example:

 There are 50 branches with many duplicate numbers: 12 x branch 0, 7 x branch 1 and so on.
 Additionally there are, at least, 3 branches missing, a 2, a 4 and a 9. Probably a 1 and a 3 as well, possibly another 2.

This all makes it virtually impossible to work out which links belong with each other.

As a further complication, rail branches can be renumbered, apparently arbitrarily, between successive updates, making it difficult to establish what, if anything, has changed on the route.

I would appreciate any assistance/clarification on this.

2

The 22/5/17 update has added a further wrinkle. There are now no duplicated branchIds in the national-rail data but, as a consequence most of the rail data is no longer present.

Note that, using my previous example, those 12 branch 0s each referred to a different route. They were not merely duplicated information.

I believe that I have devised a method of sorting out the branches into mutually referencing groups, but your latest update has removed the data that I could use to prove this.

Edit: THe update dated 25/5/17 has restored the bulk of the rail data, further analysis to follow, but there are definitely still numerous cross-referencing errors
Regards
NickP

3

Analysis of problems with bus routes (as at update 15/6/2017):

These seem to fall into two clear categories:

Firstly, routes that have genuine loops; these have unnecessary additional references,
e.g. route 244/outbound has:
branch 1 QE Hospital-Princess Alice: prev/none, next/2,3
branch 2 Princess Alice-Princess Alice: prev/1,2, next/2,3
branch 3 Princess Alice-Abbey Wood Road: prev/1,2, next/none
where it should have:
branch 1: prev/none, next/2
branch 2: prev/1, next/3
branch 3: prev/2, next/none
This affects routes 244/outbound, 265/outbound, 638/inbound, K5/outbound and R6/inbound
and also all the routes that are simple outbound-only loops: H2, H9, H10, H18 and H19

Secondly are routes with paths that diverge and then re-converge; these have a branch missing:
e.g. route R5/inbound has:
branch 0: Green Street Green-Orpington War Memorial: prev/none, next/1,2
branch 1: Orpington War Memorial-Tubbenden Lane: prev/0, next/3
branch 3: Tubbenden Lane-Orpington Station: prev/1, next/none
branch 1 is the route via the Walnuts Centre, the missing branch should be:
branch 2: Orpington War-Memorial-Tubbenden Lane: prev/0, next/3
representing the direct route, additionally branch 3 should be:
branch 3: Tubbenden Lane-Orpington Station: prev/1,2, next/none
this affects routes 88/inbound, 250/both, 358/inbound, 401/both, 485/both, 642/outbound, E7/both, R5/inbound, R8/inbound, R9/both, R10/inbound and U2/both

Other transport modes seem similarly affected but I’ll leave further analysis until the above has been looked at.

Regards
NickP

4

Is anybody looking at this problem? Been more than a month since my last post.

5

Hi @nickp thanks for your investigation of theses issues, the detailed bug reports are much appreciated. I’m raising this issue with the development team who will investigate further.

SVC-4131

6

… 3 months later … any news?

7

OK, I raised this 2 1/2 years ago and it’s still outstanding.

I have subsequently written my own routines to sort this for my purposes, using the CIF timetable files for the National Rail side and the /Route/Timetable interface for most of the rest (see separate thread for commentary on the latter)

To repeat, I believe that you have a significant problem handling routes with loops and split/rejoins. If I can make this work from basic timetable data, then so can you, particularly with the much richer data sources available to you.

Sorry to sound a bit grumpy about this, but, seriously, come on guys

8

For avoidance of doubt, over 3 years old now, not resolved by the latest update