Re: Newton and Dylan

[Jeff Dalton <jeff@aiai.edinburgh.ac.uk>]

> PC Week August 17 1992

> Its main advantage over LISP is that it is a language, not an environment, 
> and can work easily in standard operating environments (or in Newton's)."

0. Dylan is a dialect of Lisp.

1. There are a number of Lisps that are languages, not environments.
   Moreover, the move towards being a language (not an enviroment) is
   a general trend within Lisp.  For example: EuLisp, Scheme (to a
   large extent), ISLisp (the ISO dialect).

2. Lisp's problems working in standard environments are for the most
   part not due to the environmental aspects of Lisp.  Instead, they're
   due to storage management and data representation issues.

3. It seems reasonable to say InterLisp is an environment, but how
   about Common Lisp?  To a large extent, it depends on how it's
   implemented.  Common Lisp can be packaged as a shared library
   (see WCL), and (given a suitable linker) could be linked with C
   procedures rather than being an environment into which C procedures
   were loaded.

4. Of course, there are environmental aspects to Common Lisp such as
   TRACE, COMPILE, LOAD, DESCRIBE, and (some would say) SYMBOL-FUNCTION.
   Some of them (TRACE, DESCRIBE) could be defined in a language-only
   Lisp (assuming we allow an I/O library).  Others (COMPILE-FILE,
   LOAD) are there to make it possible to define systems tools in
   Lisp in an operating system independent way.  I find it hard to
   see this as a bad thing, although it's not the only good way to
   do a Lisp.

-- jeff