Activity/signal problems

[bobs]

I'm not sure if I should post this here or over with the signal threads, but I've got a problem with one activity that I can't seem to get past.

When I converted the batch of activities from their original form to work on MP3, I had a couple that I couldn't do. One was the Whitefish Local #4. It is a nice activity, so I really wanted to get it working, but I've tried a number of things and need some suggestions.

There's a pretty heavy load of eastbound traffic. There is one westbound traffic train ahead of me. I start at Essex heading west. At Red Eagle, I'm to stop and pick up two cars, then continue. So, here's the first problem. As originally written, the player path goes straight by Red Eagle, so I left it that way too. As I approached Red Eagle, I could tell from the switch indicator window that the switch at Nyack ahead of me was thrown in my favor and that I had a green signal there. I stop at Red Eagle and back up into the siding. As soon as I reverse directions, I lose the switch ahead (it is thrown normal) and now the opposing traffic has the right of way. Once I've picked up my cars and headed west, I get to Nyack and wait forever.

So, I added two reverse points to the player path. I was hoping this would tell the dispatcher that my train wasn't going to go all the way to Nyack and that he could allow opposing traffic until I've tripped the two reverse points. One allows me to back up into the siding, the other reverses the path back out of the siding again. This seemed to work just fine. One additional AI train comes by me before I reach Red Eagle, and only after I've picked up the cars is the switch at Nyack thrown for me. Just what I wanted.

OK, now, further down the road, come towards Belton. The player path stays on the main track, but has a passing path down the Belton siding. Likewise, the AI path is on the main track and has a passing path down the siding. I tested this out in the activity editor, and, I get routed on the main, the AI on the siding. He's waiting for me when I get there and takes off after I get by. There is another AI right behind him, waiting on the single track at West Belton. This train takes the siding once the first AI clears the other end, then I'm off to the races.

However, when I run the actual activity in the sim, it doesnt' work this way. As I approach Belton, I can tell that I'm being switched into the siding. That would be OK if the traffic train was on the main, but he isn't. He's waiting for me in the siding and I'm facing a double red. I tried throwing the switch manually so that it is back to normal, but the signal won't change, presumably because of the train waiting at the other end. All the switches on the route are manual.

As an alternative, I've tried having either my train or the AI train take the siding as the primary path with the other path as the passing path, but still get the same result, although with some combinations of things we deadlock on the main track instead of the siding.

Why the difference between the Activity Editor run and the real one? Why is the switch thrown into the traffic ahead of me and why won't the track clear if i throw the switch manually?

Thoughts?

[rgarber]

Bob,

I don't do the Marias route so I can't picture what you got going there. Sounds like a cool activity though. The problem with traffic trains and manual switches could be your problem. The gist is the dispatcher won't throw manual switches. So if you throw the manual switch before the AI train gets there, there's nobody to throw it back for him. It gets worse from here. As long as he's following the straight part of the switch, the AI train that is, he's fine. But no sooner he has to take the turnout, there's not much chance of the switch being thrown in his favor. One more nuance to be aware of too. When an AI train is coming from the back side of a switch, all bets are off. He'll go through because no switch needs to be thrown, so you're okay there. Like I said, I can't picture this activity because I haven't driven Marias Pass in ages.

You're next question is why does the AE do the activity right? Real simple. Because it knows you're next step is to playtest the activity. See, if it goofed, then you would know not to waste hours playtesting something that won't work, right? Clearly the AE is a conspiracy to wipe out all railroaders, world wide, so aliens can inhabit our homes and slowly TAKE OVER THE WORLLLLDDDD!!!! *cough* But I digress...

In a nutshell, the AE is lying to you... on purpose. :-)

Rich Garber

[sniper297]

The AI will throw manual switches. Once. The problem is that HAL will only set the switch for a given path one time, then forget about it. That's why the auto switches exist, they're deliberately designed so that they can't be thrown in an activity so it don't muck up the AI path. One problem with using manual switches is as Bob just described, he throws the switch and the signal don't change, but worse than that, if he pulls thru the switch onto the main part of the doubletrack, once he clears the switch behind him, it is now set for his path and against the AI train. And that's the end of that, HAL already threw that switch once, he isn't even looking at it any more. The AI train will now sit there until the switch is thrown for the curve, and HAL is gonna sit and wait for the AI train to proceed far enough to set the next switch on it's path. MSTS Standoff.
The only things I can think of for you to check, given that the signumclearahead value works for you most of the time, did you remember to check the "player drivable path" for ALL the AI services? If that's not it, a wait point or reversing point for the offending AI train might fix it.
The other option would be really painful, add another unlinked interlocking within the passing siding to space out the sequence of switch throws one more step, but then you'd have to do a route patch, the .w file, and the tdb and tit files would need to be replaced on 3.0.

[bobs]

I understand about the AI throwing a switch once, and only once, having learned about that long ago. The problem here is that he seems to be throwing it for me into the path of the traffic train. I don't want to throw it manually, and even if I do I still get a red. I get the feeling that, if I was a little slower in getting to this spot, the traffic train would be cleared through the siding and end up hitting me. I may try that out too, just to see what it does.

The really odd thing is that I recently, for the first and only time, got this activity to work all the way through, but I forced the paths around this siding so there couldn't be any possible conflict. I didn't want to leave it that way because if you end up on the siding path, you have almost 3 miles of 20 MPH!

This really is one of the better activities I've seen for MP, so I want to get it working. Another one is one of Ted Curphy's, the Columbia Falls Local part 2. I've got an idea on that one now that I haven't tried yet.

[sniper297]

Yeah, I hacked my sigcfg.dat file for signumclearahead with a number 3, which seemed to work for my purposes, then went over the rest of the route deleting all the excess signals I had in there. I've been finding out, tho, that the intervals don't trace the path down the other route after the switch throws. The AI train has to trip the next signal facing the same direction before the interval behind him turns green, the signal is still tracing the path down the siding even after the switch changes. So for V6 I gotta add another unlinked interlock to each end of the passing sidings, facing out, and test-test-test (in both directions, with traffic running both ways, are we having fun yet?) to make sure that don't screw something else up.

[bobs]

From what you wrote, I'm not sure if you understand the linking. Single head signals are always linked to the switched path, no matter what state switches are in. You can't have a signal that is unlinked. It happens by default for the singles.

I know this from looking at traces. One of the things I output on the trace is the flag that says if the linked path is active or not. For singles, this flag is always on.

On multi-headed signals, it depends on how the configuration is written.

[sniper297]

LAST EDITED ON Feb-24-02 AT 02:18AM (EST)[p]Just knew you couldn't have a discussion on AI traffic for activities without getting into the route editor problems, hey? We almost need a combined AE/RE forum for discussion of overlapping and "linked" problems (sorry, couldn't resist!). Yeah, we're kinda into semantics, you guys call them "absolutes" and "permissives", I've been calling them "interlocks" and "intervals" cuz that's what the RE object selector calls them. It still irritates me that the absolute interlockings default to that at-rest inactive stop indication when there's no active path thru them, I hate activities where the player is restricted to the exact path the activity designer layed out for him. I use the term "unlinked" interlocking to refer to a single head interlocking that's being used as an interval, I know the first head is "linked", but it really isn't linked to one specific path, so it really doesn't "interlock" anything. I generally don't use the double heads anyway, I place a single head and enable and link the second head to a switch that I want to interlock. I'm not sure what the difference is, but the gantry mounts only come that way, and enabling the second head on the pole mount effectively makes it a doublehead. So I just use the term "unlinked" to avoid confusion with one that's actually linked to a switch.

http://forums.flightsim.com/dcforum/User_files/3c78931a0fc37257.jpg
This would be a fun one to do!

[bobs]

That diagram isn't so bad. But make all of the mains bi-directional and you would have your hands full!

[Denny]

"The only things I can think of for you to check, given that the signumclearahead value works for you most of the time, did you remember to check the "player drivable path" for ALL the AI services?"
Sniper, What more do you know about "player drivable path"? I havent used that.
It appears that the routes (marias) with manual switches are fraught with their own set of problems.
thanks,
Denny
nectrackteam@aol.com
http://forums.flightsim.com/dcforum/User_files/3be893ad589dbb46.jpg

[sniper297]

One of the things that's so aggravating about this blasted game is the unpredictable results you get with everything you do. As far as I know, there is no documentation on the "player driveable path" checkbox, and the effect it has may be different on different routes depending on signals and other factors. For mine, I found that;
1. If the player path crosses/merges with the AI path, the player gets priority at the merge point, and the AI train will stop and wait at that point until the player passes, even if he's not gonna be at that particular junction for 2 hours. There are probably ways around that with the various "event triggers", but I don't even pretend to be an expert for the route editor, and I'm a rank amateur in the activity editor. So the way I got around that was to give the player a short path that starts and ends in a yard or switching district, and when he's driving on the mainline he's actually past the endpoint and off the path.

2. If the "player drivable path" box is not checked, and the player path never intersects with the AI path, every single switch on the entire AI path is set for that path at the moment the AI train spawns. On my route it takes the AI trains a little over 2 hours to travel the entire path from the starting spawn point to the end, and if I don't have the box checked, even the last switch that he won't reach for two hours is already set for him. With the box checked, the AI dispatcher (HAL9000) sets the switches a certain number of signals ahead of each AI train depending on the signumclearahead value.
Mind you, the PO&N 5 is a totally different animal from the usual, the AI paths are set up to always "take other exit", so they're routed around all the manual switches in the various switching sections via the diverge path of an auto switch. Since the AI trains always go onto the "siding", the player always takes the straight track or "main", but since the AI always has priority, once the back of the player train clears the switch it throws to the curve for the AI train's path when the AI train is a few minutes out, so the player has to stop on the "main" for the AI train to go around him on the "siding". This had to be done that way due to the nature of interlocking signals, which default to red when there's no active path thru them, and the auto switches, which default to the "at rest" straight position when there's no active path. And the interlockings are set up backwards, guarding the exit from the siding, with the second head linked to the entrance to the next siding. The exit from the straight track part of the doubletrack and the entrance from single track to double track are guarded only by intervals, which are always active even when there's no path thru them. So the player always goes straight thru the auto switches and never goes thru an interlocking, the AI always takes the curve and goes thru all the interlockings, and never goes thru a manual switch, since the AI is routed around all the areas with manual switches. Except for Sundermeyer, of course, where the AI path runs right down the middle, but that was a last minute brainstorm using the curved portion of two switches for the crossover. I'm still having fun with that.
I guess the short version of all that babbling is, it works for me, but no guarantees!