A57 - PBS and Stations Bug

[Raichase]

Having tested this, and chatted with Wallyweb about this on IRC, I have decided to post it, as it (seems to be) unknown. If it is known about, hopefully the examples here can be used to fix it.

Savegame 1 - Here we have our test train (an ET-11) bearing down on a PBS block. Within this block is a signal box, that it's orders will take it through. Now, since A53, we have not had the stations-being-the-end-of-a-PBS-block-bug.

However, in this case, the train is happy to barrel straight through the signal box (normal), checking it off on it's orders list (normal), and then run straight into a one-way signal, set *against* it. This causes the train to turn around, have another lap around this part of the network, before trying again.

The thing to realise here, is it *is* reserving it's path, as it should, but it is reserving the wrong path, as set out by TT's signalling rules (a train will never choose a path blocked by a one-way signal going in the opposite way).

Now look at Savegame 2, in which the signals are reversed, and the train picks the right path this time.

What follows is a screenie and the savegame.

[Raichase]

Now, when the signals are replaced with two way signals, the train does the same action as before, but it manages to break another signalling rule - that a train will always pick a green two way signal over a red one.

What follows is a screenie, the savegame for this screenie and my TTDpatch.cfg.

Newgrfw.cfg is below:

Code: Select all

#Always On
newgrf/canalsw.grf
newgrf/elrailsw.grf
newgrf/guiw.grf
newgrf/trkfoundw.grf
newgrf/onewayw.grf
#
# Graphics By Climate
newgrf/alpinew.grf !
newgrf/nsignalsw.grf
newgrf/ussignalsw.grf !
newgrf/arcticsignalw.grf
newgrf/grassconvw.grf
#
#New Trains
newgrf/cargosetw.grf
# newgrf/dbsetxlw.grf
newgrf/dbsetxlaw.grf
newgrf/ussetw.grf !
newgrf/arcticrenewalw.grf
newgrf/pb_brmini.grf !
newgrf/cansetw.grf !
#
#New Stations
newgrf/newstatsw.grf
newgrf/usstatsw.grf
newgrf/bare_tiles-w.grf
newgrf/basic_platforms-w.grf
newgrf/pgs-country-w.grf
newgrf/pgs-freight-w.grf
newgrf/pgs-silverline-w.grf
newgrf/pgs-suburban-w.grf
#
#New RV's
newgrf/pb_hovs_bus.grf
newgrf/gmcbusw.grf !
newgrf/334AIManagerw.grf
newgrf/334longbusesNAtruncatedw.grf
#
#New Buildings
newgrf/ttrs2w.grf
newgrf/obridge1w.grf 4713
newgrf/bridgew.grf
newgrf/highhq.grf
newgrf/pb_viaduct.grf
#
#Other New Vehicles
newgrf/newshipsw.grf
newgrf/uklcw.grf
#
#New Cargo
newgrf/newcargow.grf
newgrf/paper.grf

[Patchman]

Nice bug report!

Sadly I'll have to say that this is one of those situations where PBS doesn't work well.

The problem is that the train doesn't know about its future orders, it only ever considers the current destination. So it reserves a path to the waypoint, because that's where it wants to go, but beyond the waypoint, it still wants to go to that same waypoint, because it doesn't know that it will have gone there already, and thus it chooses a more or less random path after it.

In summary, don't give your train a choice after a waypoint in a PBS block before the next signal. If there's no choice, it can't go wrong

[SHADOW-XIII]

I had similar one with 1 or 2 waypoints somwhere in the middle of crossing the problem was there were too many train crashes :/

also I would like to report that sometimes after train crash, disappearing PBS reserved rail is too fast, while crashed train wasn't cleared fully, next (especially if fast) train will crash with the part that left

about the problem that Rai posted,
maybe it reserves too much, if it's harder to reserve proper track after the signal box why not stop reserving tracks 1 tile after the waypoint ? this way it will succesfully reach the waypoint and from there will got along, no other train will reach waypoint this way