[FIXED]2.0.2b15 Undoing changes penetration status

[tatt61880]

Edit: (by tatt61880 on 2013.02.21 00:32)
Fixed in 2.1.0b4 (or earlier). Thanks
This bug can be reproduced with 2.0.2 b15 or earlier.
-Tatt
----
Since Phun.

== How to reproduce bug ==
[scene]63122[/scene]
Trying next script will increase Sim.tick and box will move. -> No problem.

Code: Select all

for(5, (n)=>{App.Step;})

(Download the scene again.)
Trying next script will also increase Sim.tick but box won't move. -> Strange!

Code: Select all

for(5, (n)=>{App.Step; App.Step; App.undo})

I guess this issue happens because undoing initializes penetration info, incorrectly.
Serializing penetration info into scene data may solve this issue.
This can be a causes of randomness. c.f. Causes of randomness

[Xray]

Whether or not the box moves with each sim.tick seems to be dependent on the "vel" value. If I set vel = [0.0,0.0] then the box does not move. Does this make sense?

[Kilinich]

intended. app.step is for debug only

[tatt61880]

intended?
I don't think so.

[Kilinich]

intended?
I don't think so.

Yes, I'm absolutely sure, Emil add it for benchmark texting and so on.
For checkpoints there is dedicated function App.AddUndoPoint

[emilk]

Trying next script will also increase Sim.tick but box won't move. -> Strange!

Code: Select all

for(5, (n)=>{App.Step; App.Step; App.undo})

You are taking two steps and undoing one. Don't act surprised that not everything is undone

[tatt61880]

Trying next script will also increase Sim.tick but box won't move. -> Strange!

Code: Select all

for(5, (n)=>{App.Step; App.Step; App.undo})

You are taking two steps and undoing one. Don't act surprised that not everything is undone

Hmm, resent Algodoo doesn't reproduce this issue!
This bug seems to be fixed.