I was just notified that my tutorials are no longer valid in 2.60, for a very unexpected reason: The FCurve actuator no longer exists.
If you have a 2.59 file, that uses FCurve actuators, those actuators get converted to Action actuators in 2.60. This would be a minor nuisance if the Action actuators played the original IPO sequences in the same exact way as the FCurve actuators did, but they don’t, and this leads to significant problems.
Since there were no deprecation warnings in 2.59, I assumed that the developers intended to make a seamless conversion to Action actuators, which would play the original IPOs in the exact same way, therefore requiring no change to existing files.
But that doesn’t seem to be the case, so I think this is a bug.
At least, that’s what I told Colin, in this video:
I made this thread to inform others of the issue, but also to ask:
Is this something that happens frequently - where things suddenly change, without a deprecation warning, or prior announcement?
No, this is, in my experience, a fairly unexpected bug. The Python change in the BGE happened over quite a long time (2.49 had the change, but also had the old method present with deprecation warnings). I believe it’s a bug, mostly.
I think I read that there’s been a few changes regarding how F-curves are handled with pulse-mode, a number of initial problems regarding F-curves in the new actuator can be fixed with two simple steps.
1). Set the priority to 1 if it’s the only action.
2). Set the positive pulse mode on the sensor and set the frequency to one.
It does seem to work in the case of moving platforms, but that’s the majority of cases seen in my games.
However a lot of people might not know to do that right away, so a few tweaks might be needed to reduce confusion.
Yup, those are bugs. The goal was to make the action actuator behave the same. Sample files clearly demonstrating behavior differences between 2.59 and 2.60 would go a long way to getting these bugs fixed.
My test files behaved the same in 2.59 and 2.60, but they weren’t exhaustive and focused more on armature actions. I was hoping this kind of stuff would be found with the release candidates (there were 2 for 2.60 I think)
And it seems that the “parent relationship” has changed?..
I’ve an old file with a moving “dynamic” object, with a parented empty that controls other objects!
In 2.59 it works quite well,but now with 2.6O the orientation of the empty has changed ( 90° in the Z axis)?!
I created a bug report in the tracker (link in my second post). Moguri created a fix, but it was incomplete, and still left two problems remaining, so I made a small change in the source to fix one of them, and I managed to somewhat isolate the second (which seems more complex).
I’m waiting for someone (Moguri?) to either address the remaining issues, or help me hunt down the problem.
Regarding the crash: I don’t know … You can do what I did, and make an official bug report.
If you do, I’ll take a look at the .blend, and try to give you a work-around at the very least.