[e-lang] In support of formalization

[Peter Van Roy]

Mark Miller wrote:

> In Scheme's history, meta-interpreters have served this same purpose 
> well. Formalizations of Scheme only happened after it was well 
> settled. I would be curious to find out to what degree these 
> formalization activities improved the explanations of Scheme.

Meta-interpreters have several problems: they do not factor language
concepts well (interpreters tend to be monolithic) and they do not
always respect time and space complexity.  For CTM we thought
carefully about this, and our conclusion is that the translation approach
is better.  It gives us a factored semantics and it make possible a
straightforward calculation of complexity (both time and space).

Peter