Inconsistent Replays

[bateleur]

On my first attempt at the level "Awash" I came up with a solution which, whilst successful, was neither quick nor elegant.

I saved the solution and posted a link on my blog asking if anyone had any better ideas.

A short while later, a friend of mine pointed out at interesting phenomenon. My solution to the level would only sometimes succeed - seemingly the game's physics isn't actually deterministic!

Solution is here: http://fantasticcontraption.com/?designId=13887

Anyone found any other examples of this?

[pacifist]

Unfortunately contraptions may act differently on different processors. So Macs vs. PCs for example.
But I would be surprised if it acted differently on the same machine.

I fails on my machine. Are you using a mac or something exotic?

[Silver]

I've noticed this and it generally only happens on levels or solutions with a large number of loose unconnected objects. At first it bugged me, but now I kinda like it. Its like one more variable you have to take into account. Just mark it up to the wind, or the tides...

[mrmaddog]

I remember this happening to me earlier, and I am almost positive the inconsistent results were on the same computer. From what I can tell though, it happens quite rarely. I assume you are using all frame based/all time based animations.. other than that, I can't think of what could go wrong.

[Senthryl]

http://fantasticcontraption.com/?designId=11392

This always fails on the first replay for me, but always succeeds on the second (yes, on the same computer!)

I'm using a 2.8 GHz Intel Core 2 Extreme.

[Sgt Superman]

http://fantasticcontraption.com/?designId=11392

This always fails on the first replay for me, but always succeeds on the second (yes, on the same computer!)

I'm using a 2.8 GHz Intel Core 2 Extreme.

Same for me, fails the first time but makes it the second time. Interesting.

[pacifist]

It is extremely useful to have a replay that reliably breaks. Thanks for posting it.

[mgatten]

I've actually seen quite a lot of machines work differently each time. Usually, they work the same way every time except the first time - it's always the first time that they do something different. But I've also seen a couple that just act differently each time. In the latter cases, it's always a physics glitch where something asplodes and the glitch manifests different results each time.

[MetallicDragon]

I think I know what it is: If you make a wood rod (dunno about water) too short, then after you play it and press stop it'll get longer again. I've seen it happen and I think I see it happening on the four balls one.

Edit: Look here: http://FantasticContraption.com/?designId=81012, each time you press start the shorter sticks "Fix" themselves and get bigger, but then that makes other ones get too short, so the next time you press start it happens again.

[weasel]

I think I know what it is: If you make a wood rod (dunno about water) too short, then after you play it and press stop it'll get longer again. I've seen it happen and I think I see it happening on the four balls one.

Edit: Look here: http://FantasticContraption.com/?designId=81012, each time you press start the shorter sticks "Fix" themselves and get bigger, but then that makes other ones get too short, so the next time you press start it happens again.

Good call, that's exactly it. You can edit chain links (or other small constructions) smaller than they can be built, and the engine must get confused and corrects for it. I'll let pacifist know.