I understand he wants to make Guile faster and closer to Racket to make Guile more useful. Chez Scheme is his performance benchmark. He thinks users care about the language features, not about the backend that’s running the program. Apparently Racket is moving towards Chez as backend. Yet he’s spending his time to optimise Guile’s backend instead of leveraging the work already done on Chez. Anyone else thinks that GNU users would benefit more if guile chez and racket were to join their efforts? (I’m no expert in scheme or compilers, am I missing summer technical challenge here?)
Racket has the benefit of primarily being a research language and the majority of software written in Racket is pure scheme. Guile on the other hand has always been an extension language for C based projects, meaning that a significant number of Guile users have custom extensions hooking into it's backend. So an architectural rewrite of Guile would not only need to guarantee backwards compatibility of it's own language, but also of all the APIs used for extensions. Maybe there is a way to do this, but in any case it would be a huge headache. With that in mind, I think Andy Wingo is making the right decisions for optimizing Guile. Although I do expect Guile to become more chez-like over time.