Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mit-eddie.UUCP
Path: utzoo!linus!decvax!genrad!mit-eddie!smh
From: smh@mit-eddie.UUCP (Steven M. Haflich)
Newsgroups: net.lang
Subject: Re: Does C have I/O, or doesn't it?
Message-ID: <687@mit-eddie.UUCP>
Date: Wed, 31-Aug-83 09:13:08 EDT
Article-I.D.: mit-eddi.687
Posted: Wed Aug 31 09:13:08 1983
Date-Received: Thu, 1-Sep-83 03:44:14 EDT
References: <425@houxq.UUCP>
Organization: MIT, Cambridge, MA
Lines: 20

An important side issue to the desirability of excluding or including
the notion of I/O in a language is ease of portability of the compiler
itself.  There is some correlation between the "size" of a language,
the size of its compilers, and ease of porting the compiler to a new
machine.  (Of course, the effects of machine architecture on the code
generator may swamp this connection!)  As silicon foundries get more
versatile, the frequency with which new machines appear and require
compiler ports may increase.  The smaller the language, all else being
equal, the easier the port.  Compare the size of your favorite C,
FORTRAN, and Pascal grammars.  I remember once being unable to yacc and
compile the V7 F77 grammar on an 11/34, but you can count the reserved
keywords in C on the fingers and toes of one "real programmer" with
some left over.

Please, no flames!  I know modern compiler writing techniques have made
porting much easier, but it is still an issue.  Remember that much of
the stdio library is in C and survives OS ports with at least some
useful code intact (e.g. printf).

		Steve Haflich, {genrad,mit-vax}!mit-ems!smh