skaller wrote:
> Some languages define application as eager (Ocaml) and some
> lazy (Haskell) but Felix does both and neither :)
>
> The basic idea is some things are eager, some lazy, but most
> are 'up to the compiler'.
>
> So actually, three significant factors seem to exist:
> side effects or not, dependence on a variable or not,
> and finally whether the function is strict.
>   
It sounds like you are talking about "lenient" evaluation. I was looking 
into this stuff a while ago. As far as I can tell, there aren't many 
papers out there about it, but this one is pretty good: 
http://www.cs.cmu.edu/~seth/lenient.pdf.

The citeseer page might get you some more info 
http://citeseer.ist.psu.edu/schauser95how.html