On Sat, 17 May 2003, Andreas Raab wrote:
*Very* cool. What are the performance implications?
For the 'regular' sends, zero.
Great. Is it too early to lobby for this to be made part of the stock VM? As you point out, it enables all kinds of wonderful experimentation. (Yes, I know MethodWrappers provides very similar possibilities, but this is just so much cleaner).
A somewhat related question I've had for a long time - why is there no primitive for invoking a CompiledMethod without first installing it in a method dictionary? To keep the interfaces consistent, this could of course be used by CompiledMethod>>run:with:in:...
It seems like this might be useful to play with some optimizations like inline caches, at the Smalltalk level. But this is hardly my area of expertise.
Avi