Since all the rage in blog posts at the moment is laziness, I thought I’d take a look at how something can be lazy in Ioke. There are several ways of doing this, but most of them are really easy. I haven’t decided exactly how the implementation in core should look like, but it would probably be something like the following.

Basically, for laziness in a dynamic language you want something that lazily evaluates itself when requested the first time, and after that always return that value. Something like this:

foo = Origin mimic
foo bar = lazy("laziness has run!" println. 42)
"helo" println
foo bar println
foo bar println

Here we create a lazy values that returns 42, and prints something to prove what is happening. The lazy-method contains the magic. Any code could be submitted here. In this case the code can lexically close over variables outside, if wanted.

How would we implement “lazy” then? Something like this would suffice:

DefaultBehavior lazy = dmacro(
  [code]
  laziness = fnx(
    result = code evaluateOn(call ground)
    cell(:laziness) become!(result)
    result
  )
)

This code creates a lexical block and returns it. That block is a closure around the argument code. When called, it will evaluate that code, and then change itself to become that result value. Most of the magic here is really in the “become!” operation. And that is how simple it is. Since Ioke tries to make things representation independent, this implementation of lazy works in basically all cases.

By Ola Bini | In: ioke | tags: ioke, laziness, programming languages. | #