18:17:33Bikedoing some extremely limited performance tests with what i have of cst-to-bir so far, by plugging it into clasp. on (lambda (x) x) cst-to-bir is a bit faster but it's marginal. on (lambda (x) (car (car (car (car (car (car (car x)))))))) the compiler with cst-to-bir is about twice as fast.
18:17:41Bikethe latter involves more structure, i suppose
18:17:58Bikeand this is including all the later phases, like code generation and all
18:20:01Bikehm, actually that might implicate inline definitions more than anything
18:21:05Bikeyeah, ok. with something without an inline definition there's less difference. so... inlining apparently slows things down quite a bit