, Jr. Member
I thought I would try to help because some aspects of the problem you are describing, I think have something in common with one concerning sub-application windows with which I have dealt—have posted on the Adobe forums, submitted bugs, and tracked the attempted ‘fix’ in the recent ‘Mega’ SDK release. However, so far I have not been able to make much progress.
First, it would be helpful if your ‘How to Test’ notes indicated which Key you want a tester to press in order to progress from load to load. Ok, so I can read your source code to find out that it is the A key, but perhaps others won’t be so motivated.
Second, for each test I have run, I am getting:
Error: Error #3669 Bad input size.
at flash.display3D.Conte3xt3D/configureBackBuffer()
Now, since I have the Flash Player 11.2 debugger running on most of my test beds, and I don’t want that to be the possible cause of the error which you are not reporting, I have tried on all of these combinations with the same result:
Snow Leopard, Safari which has the FP 11.1 release debugger.
Snow Leopard, FF8, which may have 11.2, but probably has 11.1
Windows XP, FF8 w/11.2
Windows XP, IE8, w/11.2
Windows XP Chrome, w/11.1
One or more of those would probably be comparable to your own test environment, so I guess I need help in understanding what else I might have to do to be able to replicate the problem you are wanting assistance on resolving. So, let me know.
UPDATE:
With Windows XP running Chrome with FP 11.1 instead of 11.2, I can create the ‘empty white screen, except for non-3D content’ that you were trying to describe. So, adding that information to what maddog was able to discover about ‘Rewind’ on the context menu, I think the odds are pretty good that your problem has nothing to do with the Away3D API.
I don’t have a Flash build/test environment, but I was able to decompile your StartTest.fla to look around. The way a Flash programmer loads child swfs is not quite what I do in the pure Actionscript or AIR environments, so I am not sure how the goToAndStop((), or probably more importantly, the unload() methods behave. I assume that is the world you live in and are comfortable in, so I am going to guess that, if they behave properly after a Rewind, they are OK.
What seems not OK, and what I suspect the reason is that I am able to see the error message shown above, is that you do not consistently handle the asynchronous loading of the swf files, so there is a ‘race condition’ under which the processing necessary to establish the Context3D has not completed. I think if you enforce a consistent wait for the ADDED_TO_STAGE event for each transition from Test N to Test N+1, the race condition will ‘go away’. I would look particularly at the if (stage == null) test, since that probably returns a different result at start up than it does for subsequent loads.
I hope this may help—it turns out yours is NOT the parent-child problem that I am trying to get Adobe resolution for.