>>46
I still have a problem with arbitrarily taking a language subset and calling it the language proper. For example, I've never bothered to look into manual memory allocation in LISP (who would?) but:
it may be common to have a runtime for it to be useful, there's no absolute need to have one
I take it such a thing can be done without changing the language. Here's the rub: perhaps I don't write much LISP, but I really don't think you're writing LISP if you're forced to manage memory. If I have my history straight, LISP invented GC, and LISP machines were invented to perform GC. Sadly, we haven't invented automatic compile-time GC yet, so unless you're targeting a LISP machine you are going to need some runtime support for it.
But yes, I do agree that there are proper programs that use only a subset of their language and so do not require full language support. The split hair here is that it's not LISP itself that needs no runtime support, it's a collection of LISP programs (i.e. those restricted to a runtime-free subset of LISP) which can. (Actually I disagree with that but I mention it for the sake of argument, the comment regarding GC notwithstanding.)