No subject

[norman at nose.cs.yorku.ca]

I hadn't expected Warren to forward my note directly to the list,
so perhaps I'd better fill in some of the missing content.

What I'm trying to do with the old manuals is a mix of different
sorts of historic preservation: it's interesting to be able to
produce something reasonably close to the original in appearance,
including style differences, but I am also interested just in
making the content accessible.  That means being able to render
the manual pages into troff -man on modern UNIX systems, or into
nroff /man/man0/naa in the V5 root image, and roff and whatnot;
but also into HTML because that's the right way to make text
available on the web (Postscript is not text), and certainly into
other forms I haven't thought of yet.

To describe it all in utterly pragmatic terms, I want to be able
to put all the old manuals up on the web somewhere in readable
text form (not just page images or Postscript); and to produce
manual data of authentic content and reasonably authentic style
for use with the V5 binary distribution; and to be able to to
print clear reference copies for myself, so I can pack my old
photocopies away in a safe place; and to amuse myself by running
style and diction on the different editions; and I want to be
able to do that even if I don't have a copy of roff or the
appropriate age-authentic macro package.

So the idea is to mark up the text in a sufficiently high-level
form that it can be rendered into any of the forms above (including
the ones I haven't thought of) without undo work.  I thought briefly
about using the (V7-era) -man macros as the high-level language,
and in fact much of the simple language I ended up inventing are
obviously drawn from -man (e.g. there are constructs that are
exactly .TH, .SH, and .SS spelled differently); but I wanted to
avoid the temptation just to toss in more and more troff-specific
syntax and semantics whenever some hard-to-represent construct
popped up.  (There are too many low-level constructs in the resulting
language as it is.)  I also thought about using some existing
document metalanguage like XML or YODL, but those I looked at
were far more ornate than seemed appropriate, and far too free-form;
I don't mind carrying a few medium-sized awk programs around to
render the text, but I object to having to port a language-processing
subsystem larger than the V5 kernel just so I can render V5's manual
pages.  (Never mind how large awk and troff are these days.)

There's a name I should also name here: my collaborator in California
to whom the earlier message alludes is Jennine Townsend, who has
photocopies of my photocopies from a sort of earlier collaboration.

More on this in a few days; as I said to Warren, I hope to get
a coherent sample of all this work up on the web shortly so people
can see what I'm doing in more detail and comment, but I am in
the midst of deciding whether to change jobs (it is a coincidence
that the likely job change would put me nearer the OCR setup I've
been using, but it is convenient), and in getting back into the
swing of things at my present job after being out for two weeks
to recover from having corrective maintenance on my sinuses, so
it may not happen till the weekend.

Norman Wilson

Received: (from major at localhost)
	by minnie.cs.adfa.oz.au (8.8.8/8.8.8) id FAA25486
	for pups-liszt; Tue, 14 Jul 1998 05:00:33 +1000 (EST)