Path: utzoo!attcan!uunet!oddjob!uwvax!vanvleck!uwmcsd1!leah!itsgw!sun.soe.clarkson.edu!batcomputer!cornell!uw-beaver!tektronix!teklds!daniels
From: daniels@teklds.TEK.COM (Scott Daniels)
Newsgroups: comp.std.c
Subject: Re: Trigraphs.
Summary: A reason for trigraphs
Keywords: trigraph
Message-ID: <3603@teklds.TEK.COM>
Date: 17 Jun 88 02:35:21 GMT
References: <19345@watmath.waterloo.edu>
Reply-To: daniels@teklds.UUCP (Scott Daniels)
Followup-To: comp.std.c
Distribution: comp
Organization: Tektronix, Inc., Beaverton, OR.
Lines: 22

I was at the meeting in which trigraphs first came in, and the reason
I voted for them is fairly simple.  They were not presented as a means for
people who actually wrote C source to deal with missing characters, but as
a means for mechanical translators to pass un-encodable characters.  The
setup I imagine actually being used is:

	Programmers in ??land whose  national character set uses { for the
all-important qz ligature  (and who write comments using this a lot) happen
to have a graphic on the $ character which looks just fine as an open brace.
The programmers code away in this format locally, and (having hacked their
C compiler) everything works out.  When they decide to port their code to
another country, they can mechanically translate those chars to the proper
trigraph, and thus (1) mail source code, and (2) rely on the destination to
use their best guess for those characters.

	It was considered a great advantage by many that the trigraphs 
chosen were ugly: this meant that nobody would be tempted to write with
them, they were only for mechanical translations (a sort of least-common-
denominator format).

Scott Daniels	(I was only briefly on the committee, another startup died)
		-daniels@teklds.TEK.COM (or @teklds.UUCP)