sexpressions, sexpr, sex!

garth zenie
Tue, 20 Jun 1995 18:23:23 -0400

On Tue, 20 Jun 1995 16:58:16 -0400 (EDT)  Frank DiCostanzo wrote:

> On Tue, 20 Jun 1995, Rainer Blome wrote:
> i think i'm out of this thread except to say that i'd always heard that
> s-expression was short for 'symbolic expression' (i.e.  a sequence of
> symbols) which doesn't have a great deal to do with representation at all. 

let me try to summerize my understanding of the argument and its

fare says he doesn't like sexprs because they are defined in terms of
low level constructs.

frank says that sexprs actually aren't defined in terms of low level
constructs, it just happens to be the the most come place where sexprs
come into use is in LISP where they are in fact - defined in low level
terms (linked list).

frank is arguing for the same thing that i am arguing for (and which i
believe fare agrees with) - that there has to a well defined, simple,
easily modifiable (during runtime) syntax with which to define the
base language.  of course this would be modifiable as the parser is an
object in the language and can thus be modified or "subclassed" to
allow for derivative syntaxes.

frank says (and i agree with him) that sexprs are the best available
option.  we don't mean sexprs as defined in LISP.  sexprs don't even
have to have parens.  that's not part of the theortical definition of
a sexprs.  they certainly don't have to be stored in lists.  this is
an sexpr:


so is this of course (print (add 20 30) newline)

unless somebody can come up with a syntax which is:
	a) simpler
	b) more easily modifiable

there is no reason to adopt a new syntax.

frank, myself and some others working on misa spent a fair amount of
time working on syntax and couldn't come up with anything better than
sexprs.  if one of you can, we would be very happy to see it!

-- garth