sebj post=64617 Wrote:fritsch,
have no idea what you did to the build or this is circumstatial evidence, but I can stop livetv with your build.
Hehe. Pretty easy. I programmed the original bug, while rewriting Audio Engine. RC3 is the first release, where the Audio Device is not constantly kept open, but released after 10 seconds. So emulators, webbrowser, etc. should now work with audio quite fine (Obviously without menu sounds - as the last "click sound" would open it again for 10 seconds). And while implementing this, I changed the external calls to AE, to be non blocking, e.g. an external application that calls AE does not have to wait forever. Here I forgot basically one line, setting the masterstream NULL after deleting the memory it has. So the outputstage crash cause of this dangeling pointer :-) This is fixed now (in master).
Sad thing is, that my approx 20 testers could not reproduce the STOP in advance, so this made it into RC3. Yesterday I found a away to reproduce on my dev system and could fix it within minutes.