Re: First-class environments

I've been interested in this for some time now (but haven't spent enough
time on it).

There's two papers on symmetrical lisp that you might find interesting,
although i think they take it a little too far.

http://citeseer.ist.psu.edu/gelernter87environments.html
http://citeseer.ist.psu.edu/gelernter87environments.html

sRp

On Wed, Apr 05, 2006 at 11:28:31AM -0700, Michael Vanier wrote:

> This list has been pretty moribund for a while, so I'd like to introduce a new topic.  I've been getting 
> interested in the notion of languages that support first-class environments i.e. environments as a data structure 
> that can be grabbed and manipulated by a programmer.  The impetus for this was reading about metaprogramming in 
> Ruby, which in that language amounts to being able to extend object environments at runtime.  This makes certain 
> useful coding idioms possible e.g. reading from a database and creating objects that contain the database columns 
> as fields.  This is used to good effect in web frameworks like Ruby on Rails, purportedly drastically reducing 
> the amount of boilerplate code relative to similar frameworks in e.g. Java.
> 
> More generally, some authors have pointed out that many language constructs e.g. modules, classes, objects are to 
> a large extent just variations on the environment concept, and functions are largely just code + environment.  So 
> the appeal of first-class environments conceptually is that they might allow you to have a more minimal core 
> language in which you could define whatever module/class/object/function semantics you wanted.
> 
> What I'd like to know are: 1) What are the best/most important/most famous papers on this subject?  2) What are 
> peoples' opinions on this?  I'd especially like to hear about elegant coding idioms using first-class 
> environments and also horror stories (if any).  For instance, are there certain important language features or 
> optimizations that first-class environments make impossible?
> 
> Mike
> 
>