The original and current justification for developing MoarVM is to run Perl 6. I didn't mean to suggest otherwise.
That said, now you've raised the question of the point of MoarVM, I'm not comfortable leaving the foregoing as the complete answer because imo MoarVM is relevant to P5, ignoring P6.
First, note that there's a "P5interop" project that is starting out by embedding libperl in MoarVM. This will allow Perl 5 code to call "Perl 6 code" and vice-versa. (Sept 2014 update: See the new Inline::Perl5. This embeds libperl, is known to work with Rakudo HEAD on MoarVM, and is expected to work with other backends in due course.)
Second, note that "Perl 6 code" can include any code from any language compiled using a compiler written in Perl 6. If someone wrote a python compiler in P6, python code could call P5 code and vice-versa. In this scenario, P6 would be a behind-the-scenes implementation detail.
Third, note that there's a "v5" project, a new implementation of "pure" Perl 5. The "pure" means no XS. But don't forget the P5interop project mentioned above.
In summary, to more fully answer your question given my own personal view, the original and current justification for MoarVM development was and is to run Perl 6, but the outcome is set to include a reimplementation of Perl 5 with Perl 6 as an implementation detail (or as a side-benefit, if someone prefers to see it as such).
|