On Thu, 20 Sep 2007, Miller Puckette wrote:
It's possible to crash Pd using "dynamic editing" (deleting objects reentrantly)
That might be why I want to use reference counting everywhere.
and I don't know how to get around the problem without sacrificing run-time efficiency.
Actually, even if I don't use reference counting, the deletion operation could still scan the pd_stack to figure out whether the object is used or not. My benchmarks indicate that maintaining pd_stack (without the profiler) doesn't use much cpu (it's almost neglectible).
However, the reference counting would be a more complete solution, if I can figure out a way to do it efficiently and with enough backwards-compatibility.
_ _ __ ___ _____ ________ _____________ _____________________ ... | Mathieu Bouchard - tél:+1.514.383.3801, Montréal QC Canada