[Haskell-cafe] Preventing sharing

[Tom Ellis]

On Sat, Dec 26, 2015 at 12:26:30AM +0900, Oleg wrote:
> Kim-Ee Yeoh wrote:
> > However, the record remains that Oleg has offered little by way of elegant
> > bolting. His lazy programs based on a strict language tend to be cluttered
> > with lazy and force functions that uglify previously elegant code.
> >
> > His arguments would persuade many more folks if, for instance, he could
> > offer lazy-over-strict translations of Doug McIlroy's power serious
> > one-liners with no loss in elegance:
> >
> > http://www.cs.dartmouth.edu/~doug/powser.html
> 
[...]
> can we re-write Doug McIlroy's code in a strict language without ever
> using lazy and force or thunks everywhere -- basically maintaining the
> same structure of the code.
> 
> I took the challenge and re-wrote the powser code in OCaml, a strict
> language.
[...]
> There is no lazy/force in sight. Lazy/force could be used to implement
> the infinite stream data type (what has been prefixed with I) but no
> user of the stream library needs to know how exactly it is
> implemented. In particular, all McIlroy's code is written without any
> use of lazy/force. His code in OCaml has essentially the same
> structure as Haskell code, considering the syntactic differences
> between the two languages.
[...]
> The point is that well-chosen combinators are far more important than
> strict/lazy evaluation differences.

Thanks for that, Oleg.  I am very much in agreement with you that we can
have access to laziness within strict languages with no less elegance than
we have access to IO in pure languages.

Tom