[Haskell-cafe] Preventing sharing

[Oleg]

Henk-Jan van Tuyl wrote:
> In paper "Why Functional Programming Matters"[0], John Hughes shows how
> lazy functional programming can be used for better modularity. A more
> precise title for the paper would be "Why Lazy Functional Programming
> Matters".

The first paragraph of our paper (published 3 years ago)
  http://okmij.org/ftp/ftp/continuations/PPYield/index.html#introduction
is as follows

    Lazy evaluation is regarded as one of the main reasons why functional
    programming matters \cite{hughes:matters-cj}.  Lazy evaluation lets us
    write \emph{producers} and \emph{consumers} separately, whereas the
    two are inextricably intertwined in a call-by-value language.  This
    separation allows a modular style of programming, in which a variety
    of producers, consumers, and transformers can readily be ``plugged
    together.'' Lazy evaluation is also an elegant implementation of a
    form of coroutines, suspending and resuming computations based on the
    demand for values, giving us memory-efficient, incremental computation
    `for free'
    \cite{McIlroy:1999:PSP:968592.968597,Bird:1984:UCP,AG-embed}.

But do read the next paragraph and the rest of the paper, and other
articles on the web site
        http://okmij.org/ftp/continuations/PPYield/index.html

Our conclusion is that the modularity benefit of lazy evaluation can
be held without lazy evaluation, gaining the predictability of the
space and time behavior.