waiting to continue

martinvk

since 10 Aug 2002
Occasionally the AI will wait at a green signal for no obvious reason - the track is clear. If I go in and select Continue Schedule, the driver will continue to the next destination. In this case it happens to be a portal very far away. This doesn't happen at every signal along the way so there is no obvious cause.

Timing because other drivers are also needing attention?
The next destination is very far away?
 
I've had that too and reported it awhile ago. The AI need a bit of prodding to move along after driving about an hour or so, sometimes less. the signals will remain red and report no train approaching when there's a train sitting at the signal. In other cases, the AI will sit at stations and when told to continue, they finally start up and run. Pausing the simulator works sometimes and everything comes back to life with other weird things happening with scripted objects. After a few times of pausing, that no longer works. Quitting and resuming the saved session will work sort of, but the signals will either remain dark, although the driver hud bar shows green or other aspects. In some cases, the signals themselves show red and the AI ignores those as well as their schedule. This causes mayhem beyond what can be imagined with derailments and SPADS all over the route.

I worked N3V QA on this issue, and was told it's due to stuck threads. It appears that really complex operations with lots of drivers, schedules, and track work gets things knotted up.
 
Hmmm, with several portals generating lots of traffic, there are dozens of drivers on the tracks going here, there and everywhere. Makes for some fun and interesting background but if it just causes these knots, will have to cut back on the amount of AI traffic. Too bad.
 
I've had that too and reported it awhile ago. The AI need a bit of prodding to move along after driving about an hour or so, sometimes less. the signals will remain red and report no train approaching when there's a train sitting at the signal. In other cases, the AI will sit at stations and when told to continue, they finally start up and run. Pausing the simulator works sometimes and everything comes back to life with other weird things happening with scripted objects. After a few times of pausing, that no longer works. Quitting and resuming the saved session will work sort of, but the signals will either remain dark, although the driver hud bar shows green or other aspects. In some cases, the signals themselves show red and the AI ignores those as well as their schedule. This causes mayhem beyond what can be imagined with derailments and SPADS all over the route.

I worked N3V QA on this issue, and was told it's due to stuck threads. It appears that really complex operations with lots of drivers, schedules, and track work gets things knotted up.

Actually it can happen with very uncomplex operations, such as 2 or 3 AI trains with basic functions...……...I also reported this long long ago....
 
Some of the time on the route I use, somewhere in the max 1.5 hour session the train will set at a "false" red for several minutes. Then, eventually resume its run. I assume that it is a dispatcher in the computer control room taking a "potty" break. Hold down the shift key for a few seconds and it speeds up the clock but with a danger that you encounter some secret AI code that ain't ready and the session locks up.

I wonder if the AI will ever work without so many issues. Probably only when marketing consultants tell N3V that they are loosing "X" customers because of the poor AI performance or all of the negative commentary on the Web.

In the meantime the TANE route I am using works very well (new vegetation, track and textures) in TRS19, except for the mentioned signaling issues. I made NO changes to the signaling which is BNSF based. The usual passenger station problems crop up but do not lockup, and the short several second delays at false red signals (only one 1minute delay) are the usual things that occur during a 1.5 hour session. It takes a mysterious combination of events to be so lucky.
 
I wonder if the AI will ever work without so many issues. Probably only when marketing consultants tell N3V that they are loosing "X" customers because of the poor AI performance or all of the negative commentary on the Web.

AI in Trainz will never be fixed. We will be flying around in little mini helicopter cars before N3V ever decides to fix AI. I've given up so hard on AI. As other poster said. You can make a test layout in 5 mins with simple track and signals. AI will always, always find a way to screw it self. Simple to Complex. It's just FUBAR.
 
Last edited:
There's also the fact that, without the end user possibly knowing it, the scripts which drive much of the signalling have been obsoleted and broken by N3V updates. GWR junction semaphores are an example, where only those with a Trainz build of 2.9 or higher work (sort of).

As I posted a little while ago, Trainz with TRS19 has become a dysfunctional mess - so many different versions of the same programme, so much that doesn't work properly if at all despite the much vaunted squeals of maintaining backward compatibility.

Really time for N3V to draw a line under what is there now and start over with something fresh that will sing from the same hymn book and should be maintained as such.
 
Back
Top