Conference turris::languages

111.0. "Backus's Functional Programming system" by TLE::FELDMAN (LSE, zealously) Tue Oct 14 1986 19:48

    This is in response to 103.32, where John Bishop asks about Backus's
    FP (functional programming) language.
    
    Most of the work on FP has been presented at the ACM Symposia on Lisp
    and Functional Programming, which are held every other year (I believe
    jointly between SIGPLAN and SIGACT, but I'm not sure).  Additional
    work has appeared in the more mathematically oriented journals;
    I'd have to look things up to get exact citations.
    
    While there are clearly some similarities between FP and APL, I'm
    not sure that I'd describe FP as a "generalization of the clever
    parts of APL"; too many of the clever parts were left out.  Rather,
    it's a formalization that incorporates one of the clever parts,
    in the guise of functional forms (analogous to APL's dot product,
    reduction, and other operations that operate on operators), as well
    as the borrowing of the attitude of APL towards operating on aggregates
    of data.
    
    Unfortunately, FP omitted all of APL's neat syntax with respect
    to arrays; furthermore, there's no particularly good convention
    for record aggregates.  As a result, programming in FP can be a
    real bitch at times.
    
    Furthermore, FP still doesn't treat functions and functional forms
    as first class citizens (i. e. syntactically similar to objects).
    I believe that it is deficient in this respect, particularly when
    compared to LISP, or even to the functional notation used in lattice
    theory.
    
    I believe that FP will continue to be the subject of some research,
    but it will continue to be a minor player in this field until it
    matures some more.  In the meantime, object oriented programming
    and logic programming are likely to attract much more attention
    for the next few years.
    
       Gary