Bug Report: Passengers choose the slower connection

[chs]

I have two possible train connections between Pidleigh and Forwater.
1. A "U-turn" connection, where the above cities are located where the U begins and ends. 80 min travel time. Free capacity about 25%
2. A direct connection. 40 min travel time. Free capacity >>25%.
Both lines have about the same frequency (14/15 trains a months).

Now passengers travelling from Pidleigh to Forwater should always choose connection 2, since its twice as fast. But, they also use connection 1.
Example: train 161, left Pidleigh on connection 1, transporting passengers for Forwater.

Another example is the connection between Pidleigh and Kiling, following connections
1. same as above, Kiling is the first stop on the U-turn
3. a local train with two stops between Pidleigh and Kiling.

I found it because there was almost no free capacity on the local train 3, because there are passengers that should take connection 1 (though the time difference here is much smaller than in the first example).

The savegame "Forwater 33" is uploaded at www.mydrive.ch (public@chs78, pwd public).
There is also Forwater 32a which was saved just before I installed v10.10 (I think I had 10.9 before).

EDIT: Pak128.Britain-Ex-0.8.3

[jamespetts]

Thank you for the report - may I ask which pakset that you used for this?

[chs]

I used Pak128.Britain-Ex-0.8.3

[Carl]

I'm certain that this is a similar bug to this: http://forum.simutrans.com/index.php?topic=9262.0

If so, it's fixed by my patch in that topic. Unless the journey time of connection 1 is close to that of connection 2 (or unless they've been waiting too long), pax will always choose the latter.

[chs]

Yes, it is the same bug in that topic.

I am using the windows binaries, so I think I have to wait for the next version that includes your patch.
Its great to see that when I find a bug, others have already noticed and fixed it. Excellent work!

[chs]

With v10.11 and pak128.britain exp 0.8.3, I still get some behaviour that does not make sense to me.

1) I have a circle connection around Pidleigh, in the south east corner of the map. Train schedule is mirrored and includes the stations: Pidleigh - Berryington - other cities - Kiling - Pidleigh.
Now Passengers from Pidleigh to Berryington should board a train that goes there directly. Instead, they also board a train from pidlegh that is on reverse route and first goes to Kiling, through all the other destinations, and therefore takes the long way to pidleigh. They should only, always that those trains that are not on reverse route.
Examples: Train 267, 257, for the opposite direction 272. Train line "Z pidleigh SW Circle"

2) There is an direct connection between chilley and woolmead (Z Sackdale - Chilley - Woolmead), and a local line with 3 stops in between (ZN Chilley - Woolmead). The direct connection is faster and there are more trains on it (I considered the extra distance to sackdale). However, a lot of passengers choose the local train from chilley to woolmead.
Example: local trains 243, 244 (run it a few secs after loading, until they leave stations)

Its interesting that this seems not to happen between Pidleigh and Kiling, where the time difference is probably bigger  due to slower trains (local line "ZV Pidleigh - Kiling").

1) seems like a bug to me (or maybe a not yet implemented feature if the passengers cannot distinguish which way to go if both go to their destination).
2) may be some finetuning with the parameters carlbaker implemented (the waiting time is low)

Savegame Forwater 61 is uploaded (www.mydrive.ch, user public@chs78, pwd public)

Other than that, I enjoy the game more and more, as it gets more challenging to find optimal routes. I particularly like that overall passengers transported is now calculated more realistically, showing the potential for improvement.

[Carl]

Note that the patch discussed above has not yet been implemented in an official release.

However, the problem with circular lines may be distinct to the issues addressed in the patch.

Bug 2, however, will be fixed in the above patch -- assuming that the local train is significantly slower than the express. What counts as "significantly slower" depends on how long the connection is. For shorter hops, passengers will tolerate a greater percentage of surplus time, and vice versa.

[chs]

When will the patch be implemented in an official release?
Is there any way I can test the patch in this game, considering that I just use the windows binaries?

[jamespetts]

Carl's patches should be implemented in the next major release. If you would like to test them, you will need to be able to compile from source for the moment - they are all integrated into the -devel branch of my Github repository.

[jamespetts]

Can anyone confirm whether Carl's patches have fixed this?

[Carl]

The very first example in the first post won't occur any more, since 80 mins is not within the code's acceptable tolerance for a 40 minute journey. I understand that for the more general issues expressed here you'll want feedback from someone who isn't me!

[chs]

The general problem was split down into two particular situations described in reply 5).

Issue 1) from reply 5: This is the same as I described in a new thread http://forum.simutrans.com/index.php?topic=10752.0, it seems not yet fixed.

Issue 2) from reply 5 seems fixed, as I did not observe this anymore.

I suggest we consider the topics in this thread as closed and continue with the above linked thread 10752.