Path: utzoo!yunexus!geac!syntron!jtsv16!uunet!lll-winken!lll-tis!ames!mailrus!tut.cis.ohio-state.edu!bloom-beacon!mit-eddie!uw-beaver!uw-june!uw-entropy!dataio!bright
From: bright@Data-IO.COM (Walter Bright)
Newsgroups: comp.lang.c
Subject: Re: Efficient coding considered harmful?
Message-ID: <1732@dataio.Data-IO.COM>
Date: 31 Oct 88 19:30:47 GMT
Article-I.D.: dataio.1732
References: <3105@hubcap.UUCP> <34112@XAIT.XEROX.COM> <1700@dataio.Data-IO.COM> <8775@smoke.BRL.MIL> <7421@ihlpl.ATT.COM>
Reply-To: bright@dataio.Data-IO.COM (Walter Bright)
Organization: Data I/O Corporation; Redmond, WA
Lines: 13

In article <7421@ihlpl.ATT.COM> knudsen@ihlpl.ATT.COM (Knudsen) writes:
>In article <8775@smoke.BRL.MIL>, gwyn@smoke.BRL.MIL (Doug Gwyn ) writes:
>> (But don't write something like
>> 	for (p = &a[j]; --p >= a; )
>> which is nonportable.)
>Why not portable?

If a was created by a call to malloc, and (sizeof(a[0])>2), then on the 8086
in large memory model, p can NEVER be less than a! The trouble results from
malloc creating a segment for a. Decrementing p past a causes a segment
wrap, exactly like decrementing an unsigned variable past 0.

This behavior is allowed for by ANSI C, and is the most reasonable way
of doing things on the 8086.