A scenario changed the way it works ?

[MisterWho]

I know scenario writers can make scenarios do some complex activities, but I now have a default scenario which is driving me crazy. Since I got RW2 in October, I've been keen on "chasing yellows" and had become quite good at it, managing to score a perfect run almost every time since Christmas. Well, perfect in as much as the after-action-report had no complaints at all. Now that's all changed - every time I try this scenario, I get killed at Twyford by an HST that's sitting there waiting to ambush me. It doesn't matter whether I arrive early or late, I get an AI collision with this Oxford-Paddington HST every single time. Even when I drove very slowly in the hopes that this AI would not be waiting at Twyford, as I approached the station but while still about two miles out, it hit me even though there was no other train in sight on either track ...

So my question is whether this particular scenario behaves differently depending on the date ? Or, why suddenly for the last week is there an AI train which is behaving differently ?

Possibly pertinent information - Win7Pro 64, AMD 1090T, HD5870, Catalyst 10.10, 8 Gb memory, Velociraptor drives, onboard sound, and I clear the blueprint cache every time I start RW2. The last time that Steam / RW2 was online was January 4th when I downloaded Falmouth, and I did a verify on January 7th. Even restoring an RW2 image or copying the whole install from a backup makes no difference. Apart from the change in behaviour, the only thing that I know is different from the way the PC was at the beginning of the month is that I've applied this month's M$ security updates, but that couldn't possible be the cause, could it ?

Any help in understanding this peculiar situation would be most gratefully received,
thanks in advance,
moi

[kreader]

Could be that the scenario has become corrupted, have you tried downloading it and re-installing? Another suggestion is to go into the scenario editor and load up "Chasing Yellows" and have a look what is causing the problem. With the recent updates to the scenario editor it is much easier to spot trains that are in conflict than it used to be. There are scenario writing manuals that can be downloaded from Railsimular.com that will assist you on using the editor and if you spend some time playing around with the editor you may be able to produce your own scenario in the future.

[Kromaatikse]

When the "AI Collision" message appears, it means that two AI trains have run into each other, not that one has run into you - if the latter happens, you get "Hit Box Collision" instead.

Causes of AI collisions include one train being unable to find a destination and thus sitting idle and unnoticed on the track, or two AI trains starting off in the same signal block. Both of these are flagged by the scenario editor, but subtle changes can occur to the route which break scenarios in this way.

[MisterWho]

Thanks very much for the input. Your explanation of the "AI Collision" message ties in with the fact that this HST always seems to be just sitting at Twyford, even when I try to arrive early or late. On the other hand, when I'm early and the collision happens as I leave Twyford, there are no other trains going in the other direction which could collide with this HST. Perhaps I'm wrong in assuming that it's this HST at Twyford that is having the collision, especially as it seems that it's not a case of it colliding with me.

Regardless of which train is colliding with another, I'd still like to know why the situation has changed. I am inclined to the idea that the scenario has become corrupted, but how and by what ? Is it conceivable that the cause is the security updates that have been applied ? Would this be enough to affect the sequence of resource loading / storage such that the scenario behaves differently ? Two more tests come to mind - another verify and a revert to an o/s image without those updates. The problem with the verify is that I may not be able to tell whether any of these files have been corrected, as it will invoke the latest "updates" from RS.com - need to think about that one,
thanks again,
moi

[Kromaatikse]

If it persists after verifying Steam cache and deleting Railworks cache, the next stage is to see if anyone else can reproduce it.

Also, you might notice some coordinates in the error message. If you write those down and plug them into the navigation box in the editor, it should reveal where the collision occurs. Most likely there will be a stalled train there even before starting to play.

[MisterWho]

Weeelll, a rather frustrating day overall. Rolled the o/s back to the date of the last RW2 update / verify - crash; no difference. Tried to make a note of the coordinates, but the message disappeared before I could do so, despite the fact that it had an OK button which I didn't touch.

Went online and did a verify - no problems. That triggered an update, despite the fact that I have it set for no automatic updates. Did another verify - why ? No idea, just seemed like a good idea at the time, though since it was after 4 p.m. by then, it took 45 to 50 minutes as usual - again, no problems. Made a backup of the updated files, and tried the scenario again.

Quite a bit more AI running in the opposite direction, including more HSTs (as well as those parked at stations) and, crucially, on the approach to Twyford, an 8-car 166 which had just left. Then, surprise, no HST at Twyford, followed by ... NO crash, and shortly afterwards a completed scenario.

Why ? Just what is going on with this software ? I did two "verify integrity" runs, both of which reported all the files were OK, which implies that the only files which were downloaded were those in the update. Yet before this, for no apparent reason the scenario became corrupted, and after this for no apparent reason the scenario became uncorrupted. And despite all my efforts and your help, I'm no wiser about what became corrupted or how it happened ...

Yes, I am actually happy that it works again, but I'm left wondering when (not if) it will all go wrong again, and of course I have no idea how to fix it when it does, other than to hope and pray that a verify will have the same miraculous result as it did today.

Thanks again for your input,
moi

[MisterWho]

I may have managed to finish the scenario on Wednesday but that doesn't apparently mean the problem has been fixed. I booted up the PC yesterday and ran the scenario again, only to have it crash. Compared the backup files I took just before it ran OK on Wednesday with the files as they were then, and the only differences were in the blueprints. Rebooted, went online and did another verify cache - all OK but the scenario crashed. Rebooted, did another verify cache and the scenario crashed yet again. Rebooted, did another verify cache and the scenario finished OK. Rebooted, ran the scenario again and it crashed.

Unless someone can offer another explanation, I have to assume that this scenario either runs or crashes as a result of the way it uses resources. Even so, why would it run without any problems for over a month from early December to mid-January before it starts misbehaving randomly ? Would a complete re-install help, bearing in mind that that takes a good 5 to 6 hours even in the early morning ?

FWIW, I managed to get the details from the crash message - train 2P72 crashed at -0.47198, 51.51001. These details are the same every time the scenario crashes.

Thanks in advance for any ideas,
moi

[Kromaatikse]

You are't running Alchemy, are you? That seems to randomly corrupt what RW is doing in subtle ways.

[MisterWho]

No, neither Alchemy nor EAX in use here ...

[Kromaatikse]

Those coordinates point to West Drayton, where train 2P69 is placed. It is due to head east and exit via a portal by the time you get to Twyford at 18:30 or so. Train 2P72 is due to follow it from Taplow, and to pass West Drayton at about 18:30. I don't see any reason why 2P69 should get stuck, but if it does, that is the only way a collision at that point could occur (and then only if 2P72 ignores it's presence).

However, such random events are not unknown in Railworks. There are a large number of AI trains in this scenario, so it is feasible for memory-starvation to occur on some systems.

[MisterWho]

One thing I've noticed in all the times I've now played this scenario is that, if it's going to crash, there will be HSTs parked at the 7th, 9th, 11th and 14th stations, counting from Paddington. You pass only one other HST which happens to be moving slowly, just after the 5th station (I think), and, of course, the one parked just outside Paddington waiting for you to clear.

Contrarily, if it's going to work, then you'll pass these HSTs at speed and, usually, not at any station. You'll also pass an 8-car 166 around 2 miles before you reach Twyford. All of these movements are on the adjacent fast line, opposite direction of course.

I assume that I never see this 166 in an impending crash situation because of all the parked HSTs. Could it be that the crash occurs because of all the stacked trains (at least 5), causing 2P72 to ignore the presence of the other train ? And, as you infer, why should 2P69 be stuck in the first place ?

Thanks again for your input here. I assume that you extracted the info in your last post through RW-Tools somehow - perhaps you could give this new boy a few pointers on that,
regs,
moi

[Kromaatikse]

No, just through the Scenario Editor. Best advice there is to get in and explore.

It could well be that a HST is the ultimate problem with everything else stacking up behind it, but why would a HST interfere with a train on the slow line? I think there is more likely a glitch in the track layout or a portal, which is produced only by some random event specific to your computer.

[MisterWho]

... which makes it all the more frustrating. A random event that happens most of the time, but not all the time, and that started to happen around the middle of this month !

Thanks for the correction about Scenario Editor. Guess I'll have to find out how to get there and, as you say, start exploring ...

Thanks again,
moi

[Kromaatikse]

It could be that your computer has developed a small fault, which might be revealed by one of the stress-testing applications available. The first one I would try is Memtest86+.

[MisterWho]

I sincerely hope that isn't the case, since I only finished building it in November. However, that's what I'll be doing later today ... results later,
moi