Strangeloop Programming Languages Panel

20p_languages_panel.txt

Programming Languages Panel
September 20, 2011
StrangeLoop
St. Louis, Missouri

Members:
    Dean Wampler: Moderator 
    Rich Hickey: Clojure
    Jeremy Ashkenas: CoffeeScript, NY Times
    Dr. Gerald Sussman: MIT Prof, SICP
    Allen Wirfs-Brock: Mozilla research fellow, project editor for ECMAScript 5
    Joe Palmer: F# project
    Andrei Alexandrescu: researcher, D programming language

What is the worst idea in programming languages that afflict us today, and what can we do to make money from it?
    Alexandrescu: parenthesis *joke*
    Sussman: complex syntax
             "syntactic sugar causes cancer of the semicolon" - perlis
             "Syntax without representation is tyranny."
    Hickey: unlabeled mutability
    Ashkenas: mathematics envy
    Wirfs-Brock: source code in filesystem
    Palmer: focus on code rather than data
    Alexandrescu: dynamic scoping
                  implicit def of symbols is "typo-oriented programming"
    Wampler: nulls

What concepts and constructs are the most important to express in a programming language? (Daniel Spiewak)
    Hickey: functions and data
    Wirfs-Brock: functions with data
    Palmer: functions as data
    Hickey: data as functions

How to introduce concepts of re-use and encapsulation for user-facing "programming" environments like Excel?
    Wirfs-Brock: maybe not a space we want to be, trade-offs create worst of both worlds
                 must get data out of those systems into someplace that can be accessed by professional tools
                 don't impose concept of types in these systems
    Palmer: generalized serialization formats, data/object mapping for re-use
    Sussman: data dimensions as an option
    Ashkenas: with appropriate abstractions, text is still the perfect way to represent code

How to harness mutli-core?
    Wirfs-Brock: we are already doing it
    Alexandrescu: no clear answer yet
                  highly specialized kernels: nlp, ml
    Sussman: mechanism for interconnecting these modules is not a solved problem
             need a dynamic, organic, self-organizing communication between modules
    Hickey: stop telling the computer "how" to do things (Sussman says "yep")

Which language other than your own do you wish you had created?
    Alexandrescu: Lisp
    Palmer: COBOL, because no one else will say that *laughs*
    Wirfs-Brock: Lisp
    Ashkenas: Scheme

How to raise operating level of programmers or motivate programming in functional languages?
    Hickey: you don't solve the problem at the programmer level
            people in power must see the need and create the market
            if demand is there, programmers will commit to learning
    Sussman: break inertia of being like the person next to you
    Palmer: ask forgiveness rather than permission, sneak stuff into production
    Wirfs-Brock: environment drives adoption
                    internet drove Java adoption
                    GUI drove Smalltalk/OO adoption
    Alexandrescu: take personal responsibility
                  demonstrate productivity to peers

On the cycle of "pop culture" software concepts ...
    Wirfs-Brock: it's just fashion, cycles of human interest
    Ashkenas: it's not just fashion
    Sussman: It's not cyclical at all
             pace of progression is so fast that what we do is made quickly obsolete
             e.g. Lisp required a progression of hardware to become viable, re-spark interest
    Palmer: don't focus on language, focus on problems
            problems of the day drive language.
    Alexandrescu: spiral but not cyclical
                  progressive refinement on each pass