Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!mcsun!sunic!liuida!mikpe
From: mikpe@majestix.ida.liu.se (Mikael Pettersson)
Newsgroups: comp.lang.c++
Subject: questions about overloaded operators
Message-ID: <1346@majestix.ida.liu.se>
Date: 20 Sep 89 23:43:21 GMT
Organization: CIS Dept, Univ of Linkoping, Sweden
Lines: 39


(my statements here are made relative to C++ as implemented by cfront 1.2.1)

1) Why does C++ *require* that at least one of the operands to an
overloaded operator *must* be a class object? Does Ver. 2.0 still
enforce this restriction?

2) According to the reference manual, conversion constructors
( class X { ... X(type); }; ) are supposed to be used for conversion
of function arguments. But apparently they are *not* used to rewrite
arguments to operators (implying that operators aren't really "just"
functions operator@(...) ), making the follwing attempt fail:
	class Pair { ... };
	class Sym {
		...
		Sym(char*);
		Pair operator/(int);	// yes, I know it looks suspect
	};
	class Env {
		...
		Env operator[](Pair);	// update
	};
	foo() {
		Env e;	// empty env
		e = e["foo"/5];		// here, I'd like Sym::Sym(char*) to
					// be applied to "foo" automatically.
		...			// (since Pair operator/(char*,int)
					// is outlawed)
	}
(if you haven't guessed it, I'm trying to write environment updates in
the usual mathematical style e[name/val] rather than e.update(name,val) )

Again, does Ver 2.0 do what I want?


	/Mike
-- 
Mikael Pettersson, Dept of Comp & Info Sci, University of Linkoping, Sweden
email: mpe@ida.liu.se  or  ..!{mcvax,munnari,uunet}!enea!liuida!mpe