Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!cbatt!ucbvax!decvax!allegra!deb
From: deb@allegra.UUCP
Newsgroups: comp.lang.c++
Subject: C++ BUG (confusing error message)
Message-ID: <6490@allegra.UUCP>
Date: Mon, 23-Feb-87 17:27:49 EST
Article-I.D.: allegra.6490
Posted: Mon Feb 23 17:27:49 1987
Date-Received: Thu, 26-Feb-87 21:23:26 EST
Reply-To: deb@allegra.UUCP (David Baraff)
Organization: AT&T Bell Laboratories, Murray Hill
Lines: 47
Keywords: inline, void

/* file -- bug.c */
struct foo {
	void		doit();
	void		start();
	int		next()	{ return 13; }
	int		cond()	{ return 37;}
};

void foo::start() {}

void foo::doit()
{
	for(start(); cond(); next())
		;
}
=======================================
$ CC -c bug.c
CC +L bug.c:
"bug.c", line 15: sorry, not implemented: call of inline void function in for expression
1 error
$
=======================================

Problem: start() is NOT inline. Poor user (ME!!) went nuts looking
	 for the inline function C++ was complaining about!

Note:	 as long as start() is void (inline or no) this error message
	 results. If start() wasn't void, as next() is not, then
	 there is no problem, inline or not.

Summary: Quick fix -- make those void functions return an int a la next()
	 above. Annoying since it messes up the sense of the program since
	 this function returns a value that is ignored.

------------------
	Has anyone found this one? In switch statements, constructs
which require con/de-structors must be in a block. If they
are not, you get warned, right?

	Not always -- I had a sufficiently complicated case that C++
did NOT warn me! As a result the C code destructed something that
was never constructed -- bad news. Unfortunately, I have not saved
the code.


	David Baraff
	AT&T Bell Laboratories