An Open Letter to RSC

[Accurate]

A thousand purchases of an addon versus 100 as a result of fixing AI/Signals. The latter is more time consuming $$$. Which would you choose as the boss?

[Shortliner]

A thousand purchases of an addon versus 100 as a result of fixing AI/Signals. The latter is more time consuming $$$. Which would you choose as the boss?

I would listen to my customers. No one was asking for XBox Controller compatibility, or Steam Workshop, etc. However, everyone was asking for fixed physics, AI/Signaling and Multiplayer. It's bad business practice to ignore your customer base wishes.

[artimrj]

This letter was locked at UKTS and ts.com. Whats up with that?

[Shortliner]

This letter was locked at UKTS and ts.com. Whats up with that?

Wow, they actually allowed it to be posted at UKTS? I'm surprised!!

[bwfer]

A thousand purchases of an addon versus 100 as a result of fixing AI/Signals. The latter is more time consuming $$$. Which would you choose as the boss?

I would listen to my customers. No one was asking for XBox Controller compatibility, or Steam Workshop, etc. However, everyone was asking for fixed physics, AI/Signaling and Multiplayer. It's bad business practice to ignore your customer base wishes.

+1 Accurate!

Everyone? Everyone? ... I don't recall ever asking for "fixed physics, AI/Signaling and Multiplayer."

The TS2012 works just fine for me and I'll wait and see what TS2013 brings before I condemn it or praise it.

Factr is no matter what RSC does, it's not going to please or get of the approval of everybody. They are following the sound business model of appealing to the widest spectrum of users, and rightly so.

Brian

[philmoberg]

FWIW, I'm wondering whether it might be possible to hack an XBOX controller into an authentic control stand and brake stand. These 3D cabs are marvels of workmanship, but mouse-operated controls don't give me anything like the fine degree of control I'd have in a locomotive from the middle of the last century, much less anything newer. To the extent I can achieve a finer level of control through the keyboard, the current physics aren't all that bad (that is, they're much like what I actually experienced) with medium-length Transition Era consists. They're an order of magnitude better than they were a year ago, when a USRA Light Mike could barely budge a dozen loaded hoppers on level track. Besides, most of the feedback you actually get when running these sorts of things is through your feet, your seat and your fingertips, and I doubt it's possible to simulate that for less money than I'd have to spend for a good used car and insurance.

To be sure, there's no simulator or game program out there that couldn't stand some improvement. That said, I'm satisfied with the progress RSC are have been making, particularly over the past year. SP1 actually made it possible to run TSX on my antique graphic card, a GeForce 8400 GS, at a lightly higher level of detail than I could previously manage in legacy mode, primarily because I can now pick up the second core of my CPU. A railroad is a very complex environment to attempt to simulate, and it is an all the more complex proposition to write code for it without it's being extremely cumbersome for a system to run. To do all of that an still make payroll, while paying back the investors, adds yet another level of complication to the picture. Or would we all be better off if the investors withdrew their support?

Speaking strictly for me, if they chose to leave everything we see exactly where it is while spending the next year or so to streamline critical elements of the code, I'd be fine with that: we'd all profit in the long run. The more efficiently my current system can run the program, the less incentive there is for me to have to budget the funds to upgrade/replace it. It also means the physics and everything else will run a whole lot smoother. As long as I see steady progress in the program, I don't see anything to fuss about, even if the stuff being improved is not my highest priority.

[eyein12]

IN the scenario editor you can author a train to pick up and drop off stock( how much more proof do you need that AI have some sense of whats in the world). why cant you author an AI to pass a known consist/player using a waypoint? This does work already in game but its not perfect since you need signals and a priority, and the switches get locked in standard mode which is the major pain. however a theoretical simple new added button/marker to add in, lets say a place and time a consist is expected to arrive/wait. this would be an easy way RSC can work in an extra boost to AI functionality.

or use the select tool to drag over any of the switches that would be locked during gameplay and toggle a "player controlled" button to allow players to jump switches before an AI is set to arrive. The select tool is already in place to be used during creation of a scenario, so this added function would be in my best guess simple to put in.

this isnt rocket science, I'm a doctor not a game code writer but I already see at least 20 solutions to at least the AI issue without needing to rewrite the entire game. and that mod can be available for a small fee to offset the time it took to put it together. like I said gladly pay for it!

Ian

[SMMDigital]

It is what it is, but I take comfort in this fact: if RSC thinks WE ( the train enthusiasts ) bitch a lot, just wait until Basementboy Gamefan starts raising cain about the terrible physics that won't allow for a decent train-wreck with spectacular explosions and Grand-Theft style mayhem. Just watching a bunch of textured boxes clang together with no damage and then seeing a tiny black "game over" box pop up will get a contoller thrown at the screen real quick-like. Yes, I know the RR companies have their image to protect, but a wild-eyed troglodyte who's hoped up on Mountain Dew and Bit-O-Honeys won't care in the least about that!

[Kali]

IN the scenario editor you can author a train to pick up and drop off stock( how much more proof do you need that AI have some sense of whats in the world). why cant you author an AI to pass a known consist/player using a waypoint? This does work already in game but its not perfect since you need signals and a priority, and the switches get locked in standard mode which is the major pain. however a theoretical simple new added button/marker to add in, lets say a place and time a consist is expected to arrive/wait. this would be an easy way RSC can work in an extra boost to AI functionality.

The AI can only pick up something that hasn't moved since the scenario started. It can drop off stuff because it started the scenario with those items so it knew where they were. Anything that's moved is just an obstacle.

The AI can't go into an occupied block, and that's laid out by signal link0s. What's the practical difference between marking passing points with link0 and just having them as train order stops? the thing you can't do there is have an AI train follow you in the block, but then it won't know where you are until it hits you anyway so that's not something you want to happen.

What's a bit of a problem and might be fixable is how far ahead the dispatcher locks a route - iirc it looks four signal links ahead, although I don't know if that only counts if they're all within some distance. A little more dynamism in the dispatcher would not be so hard to do.

[Machinist]

What's a bit of a problem and might be fixable is how far ahead the dispatcher locks a route - iirc it looks four signal links ahead, although I don't know if that only counts if they're all within some distance. A little more dynamism in the dispatcher would not be so hard to do.

My experience says that dispatcher works with the concept (calcs, in fact) of "where trains will be at a certain time", pre-calculated on the timetable settings data, whatever the signal blocks ahead or distance trains are far each other. You can decieve disptacher using fictional speed of trains (%, especially of Player train).

If you set a Player to 75% (of track speed limit) chasing an AI in the next block ahead, dispatcher won't allow (you'll get the red mark, saying that trains are blocking each other), because disptacher will pre-calculate that both will be using the same block at same time. But if you set an unrealistic speed to Player (say 10% of track limit) dispatcher will think that Player will reach the block only after AI ahead has left it.

When playing the scenario Player will run normally (say 75% or even more) but that (the fictional 10%) is enough to decieve dispatcher and let you write whatever you want. Since I started to set fictional (10% or 200%) of track speed for trains (especially Player train) I never more had problem with dispatcher. Curious is that you can set 200% (or even more) of track speed limit to an AI train and it will never derail.

[Kali]

Yes, but you can tell the dispatcher one thing and do another... but it does expect trains to be where it's caculated they're going to be.

[Accurate]

It is possible to couple and uncouple several time to the same car. But, it is a real pain involving dropping a car precisely at a destination (coupler location critical) for later pickup. Then you can tell an engine to go to that destination with a pickup command and it will couple to it. Dropping a car at the precise point where the coupler is placed at the red element of a marker is the trick and not really worth it except to show it can be done. Could that be coded into the program - maybe.

Anyhow, at least the open letter may have reached the desk of RSC management by now. maybe they might say something on Facebook.

[eyein12]

you guys see my point tho. the AI functionality is really one step from complete.we have many ingenious ways of manipulating the dispatch. the only issue is the signal block and switches. its a doable fix that only needs some tweaking and added buttons. why not save it for the next release to add appeal for the next release. keep carrot in front of the rabbit.

ian

[styckx]

RSC has fallen victim to a trend so many software companies before them have fallen into and eventually go away into irrelevancy.. The constant need to feel that in order to stay relevant that they must constantly add on more crap instead of refining and stabilizing the core product and keeping the old base they promised the improvements to and at the same time impressing the new users on how stable and refined the product is. Before you know it that software that everyone once raved as showing so much promise is now a bloated unstable pain in the rump.. It's kind of ironic that the roots of the game have EA's name on it because they're pretty much following the same thought.. Instead of a new roster update, a even more broken franchise mode and some new animations you're getting some new feature you'll never play (career mode anyone?), new logos, and a even more broken physics system. (How are the rubber band couplers and exploding trains doing a year later? Fixed yet?).

Nothing is ever going to change with RSC.. They are a dog and pony show at this point and I, and so many others were conned with broken promises just so we'd keep throwing money at the screen and wondering why RSC aren't taking it any faster..

Instead of writing letters.. Stop spending your money because that letter will end up in the same place everything else that gets sent to them does.. No where.

[Chacal]

I understand RSC's reluctance to dabble with AI and physics.

Those are core functions and messing with them can break the game. Any modification would require extensive (and expensive) testing with all existing rolling stock and scenarios.

In addition, it is code inherited from Kuju. Current RSC devs may or may not have sufficient knowledge and documentation about those functions to allow modifying them.

Also the sim relies on PhysX for actual physics computation. I have witnessed, from my experience in hacking the Uru game from Cyan Worlds, how capricious Physx can be and how confusing it is for programmers that calling Physx functions with correct parameters in the way prescribed by documentation does not give expected results.

So, for a small shop like RSC, making changes to the graphics or UI, and adding new functionalities, is safe, whereas dabbling with physics and AI is not.