Where we stand

["Scott E. Fahlman" <Fahlman@C.CS.CMU.EDU>]

John,

I understand your concern about the format of the Steele book, and share
it.  I think there may be some middle ground between the current book
and "excruciatingly dull" that would allow us to produce a very precise
spec that is also useful as a manual for the working programmer (though
it would certainly not be a good introductory text for the language).
If we can do that, we will prevent the confusion that would arise due to
differences between the official spec and whatever more readable manual
everyone ends up using.

I think that the spec has two kinds of things in it: very precise
descriptions of each of the functions, forms, and built-in variables,
and some conceptual material explaining, for example, how scoping or
packages work in Common Lisp.  For the descriptions, the kind of format
you describe (and that we see coming from the Eulisp effort) is indeed
what we want.  The Lucid manual is already organized more or less this
way, which is the reason I'd like to get it as a starting point.  (We
would have to go over it carefully to get rid of any Lucid-specific
stuff and make sure it really does match Steele.)

For the descriptive material, the Steele book has some useful chunks of
text that I'd like to be able to lift.  This text has the advantage that
people have been scrutinizing it for a couple of years, so we know where
most of the ambiguities are.  If we write something new, we'll be
discovering new problems for awhile after the user community gets hold
of the new text.

I think that this mixture of formats would be good for online use.  One
could create a cross-index of function-description frames, and pointers
to the appropriate chunk of explanatory text where needed.

My inclination is to go with Tex for the new document, as this seems to
be the most widespread text-formatter around.  (I won't use any of those
evil hacks with "roff" in the name, and Scribe (TM) is too expensive for
some groups.)  If we use Tex macros in a consistent way, automated
conversion to other formatters would be just as easy as if we used raw
text in a rigid format, but the advantage is that we can quickly run off
decent-looking working documents as we go.

Anyway, that's my current thinking on the matter.  I'm still pretty
flexible on this.  Once we have access to the various sources and can
start putting this together, we can see how various organizations work
out in practice.

-- Scott