Skater teleports after cutting specific track

[jonathanolson]

In the console:

 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.0023161651783993875
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.002319885993074422
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.0022803538827247394
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.002318467059012619
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.00231471762072033
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.0023060360504132404
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.0023134035241128004
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.002328218911316071
energy-skate-park-basics_en.html:94 Changed position, wanted to change velocity, but didn't have enough to fix it..., dE=0.002285742912135902

I'll check in debugTrack 14 for testing, but:

1. I can't reproduce it with the extracted data (tapping the control point completely changes the track)
2. I accidentally hit the 'X' button instead of cut while testing later.

What happened is you could wait until the skater was to the right side of the track, then cut the upper-right control point and the skater would instantly teleport to either the ground or the (fairly far away) lower track section.

When using debugTrack with the same track info, it seems to correct itself as if it's an invalid state.

[jonathanolson]

I've created a gist with the entire console log and my attempts at inspection to create debugTrack 14 (maybe you can reproduce): https://gist.github.com/jonathanolson/2cb01de844c418bf0885

[samreid]

Not sure what is causing that. The behavior I expected is for the skater to detach from the track when it is cut, perhaps if the skater's position is being updated to the new spline coordinates before detachment, it could teleport?

[samreid]

I tried reproducing this problem a few times, but could not.

[samreid]

Perhaps in the next dev test, our QA team could check to see if they can get the skater to teleport by cutting tracks or deleting track points while the skater is on the track.

[bryo5363]

@samreid I could not reproduce this issue either. @jonathanolson Do you have any more detailed steps for reproducing the issue?

[jonathanolson]

I was not able to reproduce this either, it was very hard to trigger.

[bryo5363]

I was just able to reproduce this issue on OSX 10.9.5. It is quite difficult to reproduce. I will keep trying this method so I can develop specific instructions

[bryo5363]

I noticed that cutting one of the higher nodes while the skater is at the point of minimum PE causes the skater to immediately hit the ground. Sometimes when I cut a higher left node immediately after the skater has passed, she will flip around to the other side of the track. Not sure if this is caused by the splitting and moving of the track or just a bug.

[bryo5363]

I made a .gif so you can see what I am producing.

[samreid]

It seems important that the skater detach from the track when the track is cut, so perhaps this is an acceptable side-effect (could be more complex to apply a patch-y workaround). So perhaps we should live with this problem unless we observe that it is disrupting learning?

[ariel-phet]

Agree with @samreid here. Here is another case where as long as energy is conserved there is no issue, since it is somewhat rare use case, and we are allowed to set the "rules" of this world (ie the skater leaves the track when it is cut). If we truly feel this is an issue, we could just reset the skater to the ground and clear the energy any time a track is cut, but I believe the behavior is fine as is (since energy is conserved).

Good find regardless @bryo5363!