I have now 4 big p5p performance rants stored under ~/Documents/.
But I'm working on an official non-rant design draft https://gist.github.com/3740313 for the most common optimizing possibilities. Problem is that people are already implementing parts of it without knowing the internal problems behind.
So e.g. a function signature implementation makes no sense without lexicals const's ("is ro") being worked out before. Not talking about "is bind".
Nobody knows about perl6 and perl5i. Nobody is talking to them.
class/methods need to define the immutable/final semantics first. And I fear the decisions will be wrong. At least Stevan is a clever person not rushing things.