Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!harpo!decvax!mcnc!ecsvax!bet
From: bet@ecsvax.UUCP
Newsgroups: net.lang.c
Subject: Re: i *= f *vs* i = i * f (a partial retraction)
Message-ID: <2349@ecsvax.UUCP>
Date: Fri, 20-Apr-84 09:05:25 EST
Article-I.D.: ecsvax.2349
Posted: Fri Apr 20 09:05:25 1984
Date-Received: Sat, 21-Apr-84 02:20:40 EST
Lines: 24

Foot in mouth syndrome strikes again...
I seem to have been incorrect in stating that "i *= f" was correctly
handled by simply converting the float to int and multiplying. A number
of people have posted and mailed me notes to the contrary, some citing
Ritchie as agreeing that this is a bug.
The only really applicable item I can find in the BOOK is:

	The behavior of an expression of the form E1 op= E2 may be
	inferred by taking it as equivalent to E1 = E1 op (E2);
	however, E1 is evaluated only once.

I confess to not seeing this as incontrovertably mandating one interpretation
over another; vis-a-vis type conversion, the first clause suggests that
conversion to float for the operation and back is necessary; the second
expresses the basis for my previous stance: that assignment operators
provide a syntax for specifying two-operand instructions, for purposes
of efficiency. I still prefer the latter view; and the only really tangible
evidence I have seen offered that it is wrong is the reference to the opinion
of a demigod. C is ambiguously specified; I think this is just another example.
Lets hear it for semantic specification languages!!!

Unrepentant as ever,
					Bennett Todd
					...{decvax,ihnp4,akgua}!mcnc!ecsvax!bet