Martijn,

M> I agree it's quite the change for the current codebase... But given the
M> changes
M> involved and the idea of the monitor, isn't this what it's supposed to be
M> doing?

you see, I had this done two years ago. It required defining a virtual
method for each API function that could be hooked. At that time
Jason's idea was that calling a virtual methods instead of a static
function was too expensive if the monitor is not active. I just don't
know, as I haven't profiled anything, but I suspect that the
difference wouldn't be noticeable. I have rerouted my needs to other
solutions/workarounds since then.

Ciao
--
Nando Dessena