Hi Thomas,
that's not completely correct: the PD-like style would be to register the class (that's the actual CLR assembly) when loading it the first time. There, a static Main function could be called registering the methods. When creating the CLR object, the constructor is called, registering inlets and outlets. This has the advantage that the objects would be more lightweight (method table is needed only once per class, not per instance) and that loading is faster.
well, for that I'd use the pd-equivalent class_setup().. (that's waht external developers are used to..) ..then a setup() function for selectors and class constructor for inlet/outlets ?
WDYT ?
I've just created an xcode project where i'll track down the crashes and restructure the code. I'll propose the changes to you before comitting them.
btw, we really don't know why but with plans we noticed that in osx clr will crash if compiled against a packaged version of mono, if you install mono using fink it won't crash!
i mean the same code works well on :
- winxp
- osx with mono from fink
crash on :
- osx with mono from package
so i guess you won't find the solution in the code but maybe in the linking chain..
ciao, Davide.