[tlug] [C&C--potentially] Heisei Emperor's support of XEmacs (was: Gentoo ebuild for XEmacs-21.5)

["Stephen J. Turnbull" <stephen@example.com>]

Josh Glover writes:

 > But you're in Stanford, and it is expensive to fly there and buy you
 > beer (actually, the latter would probably cost me more than the
 > aeroplane ticket itself!).

OK, OK.  The short version is that in 1987 (IIRC) Ken'ichi Handa and
some others added the generic "Japanese patch" (supposed to cure
nicotine addiction, I hear?) to Emacs, creating ... wait for it
... "nemacs" (not jemacs, which is Per Bothner's Java
reimplementation).  Stallman quite correctly objected that (a) the
patch gave Japanese special status compared to all other non-Latin-1
languages, and (b) kinsoku processing broke English word-wrapping.

So Handa went back to work, and in 1990 or 1991 released Mule.  By
1993 it was stable and full-featured enough that he went and submitted
again.  This time Stallman rejected, again, quite correctly on the
grounds that it should be Unicode-based.  This was (a) unacceptable to
Handa on religious grounds, I believe, and (b) would have seriously
pissed off his buddies who were trying to sell the TRON encoding
(which is similar to Mule encoding, but more elaborate because it's
intended to be used as an over-the-wire format as well as internally)
to various and sundry, including the vendors in the nascent keitai
industry.

As an aside, I'll note that as of January 1995, with the Hanshin
Daishinsai, Emacs/Mule with w3.el was the only reasonable environment
for overseas Japanese on Unix workstations, as Mosaic-L10N was really
hard to build and set up, and was the only terminal-based environment
for web-browsing, period, even if you had a DOS/V PC.  It was
basically essential for web development.  There was also a Japanese
terminal app for the PC you could get for non-DOS/V PCs, but I don't
know whether it was usable with Emacs/Mule/w3.el.

At that time (1995) Lucid had already forked Lucid Emacs from GNU, and
then hooked up with Sun to support their Emacs (renamed XEmacs to be
neutral to the joint venture), and soon Ben Wing (a boy genius
consultant engaged by Sun) and Martin Buchholz (employed by Sun, IIRC)
would start to work on the Mule port of XEmacs.  By mid-May 1997
XEmacs/Mule was usable for daily work by developers or the desperate,
and by November 1997 XEmacs/Mule was stable, and the v19 branch of
XEmacs with no Mule code (even as an option) was retired.

RMS remained adamant through early 1996 (not surprising, since he
rarely admits error even when egregiously in the wrong, and in this
case he was right).  So Handa picked up Wing and Buchholz from Sun,
and the then-maintainer of XEmacs, our own Steven Baur.  Ben
telecommuted, and Martin and Steve were to work at the Mule Lab.

Curiously enough, at precisely that point in time, Handa reported on
the mule@example.com list that RMS had finally agreed to integrate Mule,
with essentially no changes.  Unicode support was not mentioned, but
Mule was to be integrated immediately.  From what I've heard, the
response on the (then-closed) Emacs dev lists was explosive, almost
mutinous, and several core developers such as Eric Naggum did indeed
depart at that time.  But RMS held firm, and 20.1 was released on
September 17, 1997 ... DOA.  20.2 was released on September 20.

I don't know that Handa or RMS has ever admitted anything, but I know
what that history looks like to me.

If you want the long (and less historically accurate but more
interesting) version, I'll need a beer.