Mon Jan 21 14:12:49 CET 2013
Faust, or point-free DSP
While I very much like the idea, my current opinion is that lambda is
king, and if one wants a combinator approach, it should be written on
top of lambda. Both named/unnamed data approaches are valid, but
writing unnamed on top of named (lambda) is easier than the other way
I also no longer see the point of writing a DSL that is not embedded
in an "old" tried-and-true functional language (Haskell or Racket).
Writing languages and compilers for DSL is hard enough without having
to also re-implement the low level details like symbol management and
That said, it might be an interesting application to implement Faust
on top of the stream language I'm writing here.