Early Lisp Part II
funcall.blogspot.com
funcall.blogspot.com
I'm not sure what to say about it, other than it's well worth your time. It's also a lot of fun to implement the interpreters it presents.
Eventually it settles on emacs-style global namespaces, after more or less proving that every other alternative doesn't really work.
The whole point of macros is to control evaluation. Sometimes you want intentional capture; it's not accidental. Throwing out the chainsaw just because a few people use it improperly means you won't be able to cut trees so easily.
But, different people can feel differently about it. :)
Hygiene is opt-out (except maybe in explicit renaming), not opt-in like in common lisp. That is, IMO, a good thing.
I would not at all be happy to not have a facility to do capturing as well, though.
Seems like the right default to me.
It's just... archaic.
That aside, I think Lisp is genuinely at a sweet spot for the product of nice-to-program-with and simple-to-implement. It's historically important besides, and there's lots of code worth reading in it (SICP and PAIP just to start).
That's one reason that I said of any dynamic language. Static languages have different tradeoffs of tooling -- you're not going to redefine a function in them anyways, but perfect static analysis for lots of things becomes almost trivial, since the compiler has all of that information available to it.
It feels like every other language started from the point of "I like lisp, but this one feature is too dangerous, let me remove that."