Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!decvax!yale-com!leichter
From: leichter@yale-com.UUCP (Jerry Leichter)
Newsgroups: net.lang.c
Subject: Re: Supporting Structure Comparison
Message-ID: <2124@yale-com.UUCP>
Date: Wed, 5-Oct-83 09:23:01 EDT
Article-I.D.: yale-com.2124
Posted: Wed Oct  5 09:23:01 1983
Date-Received: Fri, 7-Oct-83 03:59:03 EDT
References: unc.5972
Lines: 31

One more general approach - which I won't either advocate or criticise, for
the moment - has been taken by some compilers (I saw it in some HP compiler):
You allow user-defined operator extensions.  Thus, if a and b are of type
COMPLEX,

	a + b

is compiled as if it was

	SUM_COMPLEX_COMPLEX(a,b)

or some such thing.

To really make this usable for things like complex arithmetic, where I would
want the complex sum to be expanded in line, it should allow SUM_COMPLEX_COMPLEX
to be a macro.  Unfortunately, this would be difficult because of the timing
of macro expansion and expression parsing; the compiler I saw just generated
a function call.

The big question involved in such an approach is:  Can you keep it small and
simple enough to be understandable - and implementable! - while still making
it useful?  As a quick example of the kind of problem you run into, if a is
as above, what should you make of:

	a + 1

If this calls SUM_COMPLEX_int, you'll end up having to write a LOT of functions
just to get things off the ground; if it calls SUM_COMPLEX_COMPLEX, things will
get very confused.
							-- Jerry
					decvax!yale-comix!leichter leichter@yale