X-Authentication-Warning: delorie.com: mail set sender to djgpp-bounces using -f
From: "Rod Pemberton" <dont_use_email AT nohavenotit DOT com>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: _CRT0_FLAG_NULLOK
Date: Tue, 06 Aug 2013 07:34:36 -0400
Organization: Aioe.org NNTP Server
Lines: 76
Message-ID: <op.w1d7ry2o0e5s1z@->
References: <21e77579-1a40-4442-8111-fc976fba78fc AT googlegroups DOT com>
 <b5o37qFldl9U1 AT mid DOT dfncis DOT de>
 <f24e8cd3-83ca-4386-a319-a9adb4d74c68 AT googlegroups DOT com>
 <3df2f50f-9543-47a7-8e40-a9be82ce5018 AT googlegroups DOT com>
 <87fvuvny2v DOT fsf AT uwakimon DOT sk DOT tsukuba DOT ac DOT jp> <op.w020sren0e5s1z@->
 <87ehaeonbp DOT fsf AT uwakimon DOT sk DOT tsukuba DOT ac DOT jp> <op.w04oetou0e5s1z@->
 <201308011247 DOT r71ClEhM008182 AT delorie DOT com> <op.w08sn91j0e5s1z@->
 <201308031802 DOT r73I2NAq010284 AT delorie DOT com>
NNTP-Posting-Host: CNsg4fVcCsvs3UaOgZtQCw.user.speranza.aioe.org
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit
X-Complaints-To: abuse AT aioe DOT org
User-Agent: Opera Mail/12.16 (Linux)
X-Notice: Filtered by postfilter v. 0.8.2
Bytes: 4592
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

On Sat, 03 Aug 2013 14:02:23 -0400, DJ Delorie <dj AT delorie DOT com> wrote:

>> A NULL is not required to be mapped to low memory.  This is
>> irrelevant.  NULL's actual value is not required to be zero, only
>> different.
>
> NULL's actual value is required to be zero, from the program's point
> of view.
>

If it's from the program's point of view, then it's not it's actual
value.  It's its semantic representation.

>> (No joke.  You can ask even the angry nutjobs on c.l.c and get that
>> correct answer.  Please read the two posts which I provided links
>> to.  They're by one of the authors of the ANSI C specifications.)
>
> Sure.  The physical address of "0" in a virtual memory system is
> rarely itself zero, and some systems use segment/offset or tag/offset
> addressing too.
>

That's only true if paging or MMU is involved.  C was implemented
prior to virtual memory on many machines that didn't have it.  All
microprocessors started out with physically mapped memory.  Neither
the 68K series nor the x86 series had MMUs for a few generations of
each processor.  To this day, the mode x86 starts in is physically
mapped, not paged, not segmented, not virtual.  DJGPP seems to be
unique in the DOS world for using virtual memory.

The point is on *any* memory system  The actual value used for
NULL needn't be zero.  It just needs to not collide or compare
equal with the values used for other C objects.

>> True.  We were discussing how to make a NULL more useful by
>> preventing atoi() from failing when passed in.  One method is for
>> NULL to be not just a "char pointer", but also be "a pointer to a
>> string" - an empty string.
>
> IMHO, making "passing NULL" crash your app is more useful, because
> then the debugger can tell you exactly where you went wrong.  Even the
> MCU simulators I write for work have *NULL==crash for debugging.
>

So, I take it you also disagree with the C compiler automatically
initializing unitialized variables.  That prevents users from making
coding mistakes too...  Shouldn't they clear their variables themselves?
I'll also take it you dislike 'void' and 'void*' since those also
prevent users from coding making certain coding mistakes with pointers.
Wrong assumption?  Of course, it is...  I seriously can't imagine you
disliking either of those.  I.e., I don't see why you'd like the just
mentioned "safety" features of C, but you wouldn't like another change
that would prevent similar novice style coding mistakes...

> IMHO if a pointer points to *anything* that "works", it's
> a bad choice for NULL.

Even if NULL still functions correctly as NULL?

Why would that be an issue?

NULL can still be compared to.  It just now functions as an empty
string for functions expecting strings, if they don't explicitly
check for NULL.  If they do, it functions as NULL.  That can prevent
runaway on a call to strlen(), strcmp(), if the function didn't
explicitly check for NULL.  By "runaway", I mean searching for a
null character '\0' when one hasn't been properly placed.


Rod Pemberton
-- 
"... to secure these rights, governments are instituted among men,
deriving their just powers from the consent of the governed; that
whenever any form of government becomes destructive of these ends,
it is the right of the people to alter or abolish it ..."
-Thomas Jefferson, Declaration of Independence, 1776