> does more for you than the equivalent dynamically typed code. It's _not_ the same result.
Yes. It does more checking at the cost of imposing restrictions, such as forcing homogeneous containers and conditionals and having less generic, less cluttered code. This is not merely hand-waving, BTW. One just have to take a look at some decent Lisp code, such as arc.arc or some good parts of Common Lisp or Norvig's code from AIMA which is simply wonderful.
As for typing, an old-school ADTs are OK for me (that is constructors, selectors and predicates explicitly defined as procedures). This requires some discipline, because the type system does not do anything for you, but all this is trivial, including writing pattern-matching.
I could argue that strong typing via type-tagging of values (values has a type, not variables principle) is good-enough as long as it comes together with other Lisp's features, such as everything is an expression, everything is a first-class value, which gives one such beauties as the Numerical Tower, but this is quite another topic.
I am also OK with the SML-family languages, love Haskell for its clarity and conciseness and have nothing against them, but... I still think that there are prototyping languages and implementation languages, and I still prefer to prototype in a dynamic-typed language and would still use Common Lisp if I could have a choice or Clojure.
Yes. It does more checking at the cost of imposing restrictions, such as forcing homogeneous containers and conditionals and having less generic, less cluttered code. This is not merely hand-waving, BTW. One just have to take a look at some decent Lisp code, such as arc.arc or some good parts of Common Lisp or Norvig's code from AIMA which is simply wonderful.
As for typing, an old-school ADTs are OK for me (that is constructors, selectors and predicates explicitly defined as procedures). This requires some discipline, because the type system does not do anything for you, but all this is trivial, including writing pattern-matching.
I could argue that strong typing via type-tagging of values (values has a type, not variables principle) is good-enough as long as it comes together with other Lisp's features, such as everything is an expression, everything is a first-class value, which gives one such beauties as the Numerical Tower, but this is quite another topic.
I am also OK with the SML-family languages, love Haskell for its clarity and conciseness and have nothing against them, but... I still think that there are prototyping languages and implementation languages, and I still prefer to prototype in a dynamic-typed language and would still use Common Lisp if I could have a choice or Clojure.