From mboxrd@z Thu Jan  1 00:00:00 1970
From: lm@mcvoy.com (Larry McVoy)
Date: Mon, 14 May 2018 08:04:31 -0700
Subject: [TUHS] Who used *ROFF?
In-Reply-To: <F6DADFE5-62B0-4312-9668-90FD45AF1CC6@ccc.com>
References: <201805141219.w4ECJo5G030533@coolidge.cs.Dartmouth.EDU>
 <alpine.BSF.2.21.999.1805142237130.79090@aneurin.horsfall.org>
 <F6DADFE5-62B0-4312-9668-90FD45AF1CC6@ccc.com>
Message-ID: <20180514150431.GB26148@mcvoy.com>

> Ditroff was reimplemented by Clark (IIRC) to create today???s groff which takes mostly a superset of the ditroff input language.  

Yep.  In early C++ which I found questionable but he made it work.

One of the superset things is something I got him to do in pic, the
'i'th construct.  This chunk of pic:

	for i = 1 to units by 1 do {
		line <-> dashed from `i'th [].C.s - (.10, 0) to \
		    last box.nw + (i/(units+1)*w, 0)
	}

is part of the code that produces this:

	http://www.mcvoy.com/lm/sunbox.pdf

and I could change "units" and have more or less machines.  That's a diagram
of the first cluster that Sun shipped, code named sunbox, shipped as
SparcCluster I.  My baby, never went anywhere, but my product marketing
guy came up to me about a decade later, after Google was a thing, and
said "I guess you were right about that clustering idea" :)

Source for the diagram is here:

	http://www.mcvoy.com/lm/pic/sunbox.pic

Traditional troff can't handle that unless someone backported the 'i'th
construct (which is obvious, right?).

--lm