[racket] Interesting article

On Thu, Aug 12, 2010 at 4:14 AM, Deren Dohoda <deren.dohoda at gmail.com> wrote:
> 4) Well, *I* have no clue what you want, so I'll just create this view
> called a "sequence" and give you that (oh yeah it just acts like a
> list)

[...]

> they use a list in the first place?). I don't believe (4) is so much a
> "solution" as a way to relocate the problem to post-processing of your
> containter-viewed-through-a-sequence rather than pre-processing by
> selecting the right function in the first place.

It's then the compiler's job to specialise your map function for your
container. You get the benefits of conceptually treating the vector as
an abstract sequence, while concretely having your functions
specialised for your specific container, and only having to make the
implementation decision at a single point in the code.

martin