>I think it's also a very different approach, more like placeholders in
>jitlib (correct me if I'm wrong - I've never used jitlib). Loose ends
>allow you to build a program from nothing and have it active without
>being structurally complete (from my brief reading it seems that the
>program pauses if it finds a loose end, waiting for the programmer to
>write that bit).
Yes, I think that is similar. The basic idea of this type of delayed
evaluation is that programming is the composing of structures that do
calculations (like generators, or abstract functions), instead of
doing the calculations straight away. A closure (like in LISP) is
something like this already, since you can bind the arguments later.
The problem is that the whole structure must be calculated again when
you change the bindings. The proxies in jitlib are placeholders that
do this binding dynamically, so their graph can be recomposed at
runtime.
>Programming to pre-written tests is all about
>achieving some pre-defined ideal, and you always have to have a
>structurally complete program to run the tests against.
tests are great to learn to understand what the role of a certain
type of object is. But apart from this, as you say, they don't
guarantee anything.
>> I cant find LC2 on this, but it might be a good time to send it round
>> again anyway:
>> http://www.oreilly.com/pub/a/oreilly/news/languageposter_0504.html
>
>Beautiful, we could make one for dynamic languages.
I have the one from the History of Programming Languages, that goes
till 1970 (and it is pretty "networked"), so I could complement the
early history..
--
.
Received on Thu Sep 21 2006 - 12:16:58 BST