Transport Tycoon Forums

A double signalbug : train 6 will choose the wrong way at the upcoming PBS junction because train 139 "blocks" the double signal just after the depot.
Removing the double signal makes everything work again, I.e. train 6 will choose the right way again.
I however use this kind of setup to prevent lockups as the train have an escape route into the depot.

Ok, the double signal bug mentioned abvove doesn't seem to be repaired completely.
Most of the time it works but every now and then a train decides differently and just passes the brach it should take, screwing up my flow of trains.

My ttdpatch.cfg file hasn't changed since the above posting.

Well, since it's not a PBS junction, it's not a PBS problem and so due to a limitation in TTD's path finding.

Indeed it is.
I disabled pathbasedsignalling and the train still takes the wrong route.

So..many...trains..

Found another, probably old bug.

If train 73 turns around, either manually or if the signalwaittime has passed (due to a network lockup), and it reaches the the PBS exit signal in front of the tunnel it will turn around again, but then train 223 also gets the green light and will crash into train 73.

It must have been an old bug because I think most of my crashes are caused by this but I never got around to catch it, since I was to busy solving the lockup in my network.

ps if I change the PBS exit signal into a normal signal the bug doesn't happen.

Train 17 (the one waiting at the signal, headed for the depot) needs to reverse but cannot do so safely.

(Methinks having "signal2waittime 2" might be causing a few problems with my terminus stations. But it helps keep my Ro-ros from getting jammed. Maybe I need to move my depots out of the PBS blocks?)

A trick for Ro-Ros.

axb-c*************dx-
axb-c*************dx-


a = approach signal (one-way pbs presignal entry).
b = two way pbs presignal combo.
c = one way pbs presignal exit.
d = starter signal.
x = crossover
* = platform

Note there must be a reservable tile between b and c.

Using this layout, trains will pick a platform (because b is a two-way signal), but they will never try to go back the way they came (because c is a one-way signal).

I always have signal waittime set to 0. That is the most realistic and the most predictable.

Yes indeed, I invented this station style months ago, but I've never seen anyone else use it. It works quite nicely. The only trouble is that you do have to have enough room for all the signals and spacing.

No, what causes my problem is the depot. I do know enough to turn reverse-at-stations off; occasionally I wish it were on, but it solves far more problems than it causes.

 D
axb***********dx-
axb***********dx-

Same, but with D==depot.

I guess I should have said "move my depots out of the presignal blocks".

I usually have that one on.

I haven't had any real problems with depots, but I always give my trains servicing as part of their orders, so I always know which direction the trains will be trying to get into the depots from (and where they'll want to go when they leave).

I've encoutered a new/old PBS bug, sadly but true.

If a train stops with his rear end just inside the PBS block, and then tries to turn around (signalwaittime has passed) the Patch Safety Measure (PMS) kicks in and stops the train from turning incase the PBS junction behind it is occupied.
However the PMS fails in this case because the reserved track behind the stopped train is cleared by this action and the upcoming trains crash into the stopped train.

Trains crash at this station, but only when I am modifiying PBS junctions in other parts of the system.

I can't reproduce this bug on demand. But with this station setup and a lot of trains coming into and out of it, I sometimes get crashes in the entrance. I can leave the game running for (game) years and nothing bad happens. It's only when I am changing signals someplace that the crashes occur.

version info

TTDPatch V2.0.1 alpha 67 (Windows) starting.
TTDLOADW.OVL not found, looking for original files:
Copying GAMEGFX.EXE to TTDLOADW.OVL
This is the Windows executable.
The program name is:
  Transport Tycoon (Deluxe) V3.02.011 Copyright (C) 1995 Chris Sawyer
The exact version is 3.02.011  11th October 1996 17:23:14
Parsed version is V3.02.011, size 1695744
Setting program size to 8.5 MB.
Installing code loader.
TTDLOADW.OVL is OK.
This version of the program has known addresses.
Starting TTDLOADW.OVL off
Result: [OK]

I'm afraid that if it's not reproducible, I won't be able to fix it.

Note however, that this station entrance has a problem anyway: it's too big. No train will ever enter the last platform, and no train coming from the right lane will ever enter the last three platforms. The reason is that PBS paths are limited to no more than 7 junction tiles.

I doubt that's the reason for the crashes, but without being able to reproduce the crash myself I can't say more than that.

PBS is the most brillaint thing since pre-signals.



Fair enough. I spent hours and hours trying to get a clean repro with no luck. Just be aware that there may be a race condition that happens when a new PBS junction is setup someplace else.




It is just 7 tiles by my count. See screenshot. In practice this station is handling a lot of traffic with great throughput. I have an altered design that never has crashes, but it's isn't quite as effecient.



I think that trains reversing have something to do with the crashes. This station can get backed up and the trains come out of the station due to me allowing them to reverse in stations.

You're right, I counted some tiles as junction tiles that weren't.

Well, let me know if you can somehow reproduce it, then I can have a look.

I've had something similar but non-reproducible. A large station with relatively limited exit choices where trains wishing to leave the station end up reversing.

You might be able to avoid the problem by setting signalwaittimes far higher.

This is pretty much right on. I set my signalwaittime for signals 1 and 2 to 255 and the trains don't crash anymore.

After spending some time trying to isolate a reproduceable crash, I determined the problem was due to reversing trains as in the problem describe a few messages up.