I mean this as no “red herring,” Talex. I merely offer the observation that “the worthy successor to (so to speak) Perl-5” just might have been staring us all in the face for several years now. And it isn’t called .. god only knows what they call it these days .. it’s called Moose. If we turn our attention to implementing Moose within the actual perlguts, we just might obtain a timely return-on-investment.
(I hope that I am not speaking to an empty chapel here ...)
I’m not sure that performance is necessarily the big-win here: the Perl core is already so efficient that it can actually withstand Moose being grafted on top of it. To me, compile-time is the biggest potential win/driver to the idea. Unlike “regular Perl,” Moose does have the potential to introduce strong-typing. isa => 'int' might constitute “laborious syntax,” but it nevertheless could serve to associate a strong-type with an attribute ... a strong-type that could be made to have “teeth,” including before the program starts to run.
Furthermore, we will have moved out of the “pie-in-the-sky hypothetical Land Of –Er,” and back into a frame of reference that has already been implemented and put into production many thousands of it-pays-the-bills right-now times.