Corrupt Game

[jeffryfisher]

Playing my latest US History map, I have reached a point at which the game will always crash at the end of the month (when there's a long delay while it seems to be trying to build some new AI track). I am hunting for anything strange on the map (like a station whose frontage track pokes into a river).

If you guys have any ideas about other kinds of "land mines" a game can plant on a map (especially if it would screw up an AI track-build), then please chime in. I am attaching a zip if you'd like to eye-ball the map yourself.

BTW, You can also see how I play my map from starting in Mexico.

Amex3.zip

[jeffryfisher]

Well, maybe there's a math error in one of my events. I tried running from a save one whole game-year earlier, and I got the same crash in the same manner at the same month. This crash is coming at the end of the first month after Mexican RR nationalization, so there are three month-end events being triggered to calculate how much money to add/subtract from the company in order to pull it toward the government's 5% profit allowance/guarantee.

However, this same set of events has been running smoothly in other games where my US companies are nationalized during WWI. So... I'm still not sure what's happening. I do know that there's always a "AI build delay" just before the game crashes, so that could still be the problem. The AI has been remarkably consistent in repeating all of the same builds in all of the same months even running repeatedly from a saved game a whole game-year back.

BTW, If you're a techie/debugger who has my US History map, the events in question are near the bottom of the event list (in the editor). They're labeled like "Nationalization Chomp" (& Nibble & Subsidy). Looking through them, I do not see a div by zero (yet).

Sometimes I feel like a dog who's chasing a parked car and barking up the wrong tree at the same time.

[jeffryfisher]

Update: I found an older save of the same game that will play through without crashing, so my events are less likely to be the culprits. Instead, I am now looking at my track builds across the US. I wonder if I cramped an AI RR in a way that its next build crashes the game. If I can find the grind spot, then maybe I can steer the computer company away from it.

At least I proved that I can reach silver (on my way to gold) playing Mexico starting in 1844 :)

[Gwizz]

Another thing that will crash a map is covering part of a farm with a lake in the editor. Then building track across the remainder of the farm in a game.

[jeffryfisher]

I don't think that my map had any buildings whatsoever in the editor. All economic objects are dynamically generated to avoid one of our other favorite bugs.

[jeffryfisher]

Well, my analysis of past saved games failed to indicate what build "caused" a problem. Unfortunately, I did not still have the series for the big-build out of Mexico. Sooooo...

I have gone back to the save immediately prior to the building project (thank goodness that I save builds separately from operations!). It is able to run through the point at which the crash was occurring in the later saves. I am building out that whole transcontinental extensions and branches again (at least I have a finished example to remind me what decisions I made). This time, after each ~50-cell section, I am saving and testing (running ahead a year, then going back to my build-save). I have not yet recreated the crash. Maybe I never will. I'd be forever puzzled, but I wouldn't mind a clean build.

[jeffryfisher]

The mystery deepens: I have recreated nearly all of the problem rail-build and stations and new trains, and I can fast-forward way beyond the other instance's crash date without recreating the crash. Maybe there was simply a corrupt byte somewhere in the other file. Maybe there's a bad sector on my disk (I should do a scan).

The one suspicious part of the build was where I bulldozed a logging camp to squeeze my line straight through between some mountains and a river. The first time I played it, I cleared one cell of the camp (I don't remember which). This time I scoured every cell and some of the adjacent cells just to be safe.

[Gwizz]

I still have game crashes. So, like you,I save often. On the last crash I went back, one month at a time and the game crashed just as 3 trains came to a bridge crossing. just as the first two trains touched the game crashed.

I still don't think the trains caused the crashed. I believe the ME computer I use for gaming is becoming more sensitive when the computer tests too many triggers, etc. at the same time.

Also, the hourglass appears often at monthly changes since I normally have 12 or more railroads in operation but seldom crashes the game.

The computer was over clocked years ago and still seems to be faster then my 7-64 machine. Because it is a 64 machine it won't run Railroad Tycoon. My XP machine is still waiting on repairs or someone smarter then I to fix it.

Since I put the bigger fan over the CPU on the ME machine , The crashes are now few and far between. But this last crash totally stopped the game at that same repeatable spot.

I was testing win events and had one win to go. I went back about 4 month and used the cheat code (slush fund) to build net worth and the Double Gold win fired just before the trains came together.

These bugs can waste a lot if time when a game is heavily evolved with activities.

[jeffryfisher]

...the hourglass appears often at monthly changes

Month-end delays are caused by AI managers thinking about building new track. As you can imagine, track plotting is an iterative process with many complex conditions, so it can suck down some serious CPU cycles. The 400+ events in my US History scenario are nothing by comparison. Even when several of the events are doing nested loops over both territories and companies, they all run in an instant. Newer processors are so much faster than what the game was designed for that even my "aging" dual-core 2.9 GHz CPU can handle all of the events every month without trains on the map even hesitating (at game-speed 8 ).

[jeffryfisher]

Woohoo! (sort of)

I have absolutely identified the proximate cause of my crash. I don't (yet) understand WHY it causes a crash, but it's where my New England branch RR crosses the Bangor & Brunswick RR near Bangor. I narrowed down the crash to building just one cell-segment of track into the B&B from the SW. I don't know why, but for some reason, joining into that AI RR is a big NO-NO!

Without the build, I can fast-forward 3+ years. With the build (just one cell of track connecting to the AI's RR), the game always crashes about one year into the future, during a long pause between months, presumably thinking about how an AI can build new track. Who knows? Maybe my connection opens up so many possibilities that the AI overflows its stack trying to think of a way out of New England.

Anyway, now I must engineer a new path into New Brunswick Canada. First I'll try crossing the AI at right-angles to see if that avoids the bug. If that still crashes, then I'll have to take a more painful detour (or maybe just buy the offending company).

[jeffryfisher]

Well, I seem to have avoided the crash by twisting my RR into a right-angle crossing of the toxic company's track. A few game years have gone by, and that AI company still hasn't built beyond its initial Maine short-line. My best guess at the cause of the crash is that joining into that little line opened up so many possibilities that the game overflowed a stack when it tried to build from them.