John Backus's Turing Award Lecture
John Backus's Turing Award Lecture was a watershed for the
programming-language community because the inventor of FORTRAN, which
was the dominant programming language of the day, stepped forward and
said that the main stream of programming practice was flowing in a
most unproductive direction.
His lecture "Can Programming Be Liberated From the von Neumann Style?"
was developed into a
fairly technical paper
which opened the floodgates for a torrent of intellectual activity
that continues to this day.
While Backus's message remains as relevant as ever (simply substitute
"object-oriented languages" for "conventional languages"), the
technical work has worn less well.
-
Thanks in part to a lifetime of work by Richard Bird,
the
algebraic reasoning techniques in Section 12
have been developed to a point where they are much more readable and
can be routinely applied to
small problems;
they are especially useful for developing efficient
algorithmic cores.
But like other formal techniques, they are not scaled up
to write whole systems.
-
Functional programming has moved away from the point-free style
inspired by APL and shown
in Section 11 of Backus's paper.
Current functional languages deliver more expressive power in a more
readable way.
Backus's original proposals, while tremendously important to the
development of a young field, are now taken for granted.
Much of the most exciting work builds on two ideas brought to fruition
after Backus's lecture:
polymorphic type systems with type inference,
and algebraic data types
with pattern matching.