ms1g@ANDREW.CMU.EDU (Mark Steven Sherman) (06/20/86)
Enough already. One can pass fully specified procedure parameters and type check them at compile time. One can use name equivalence for the checking. The Acta Inf. article does not contradict those statements, and there are both empirical examples(I wrote a Paragon compiler that does so - which does not provide recrusive modes in Algol 68 terms) and there are formal models (check out any axiomatic description of complete Pascal -- if you can prove that a specification is met, then you had to prove that type checking was correct). Acta Informatica is available in nearly every technical library. Go read the article if you want to see what it was about. (For those still interested, Clarke's article was in POPL several years ago -- people who want the complete reference should contact me and I'll dig it up.)