Path: csiph.com!usenet.pasdenom.info!aioe.org!.POSTED!not-for-mail
From: "Rod Pemberton" <dont_use_email@xnohavenotit.cnm>
Newsgroups: comp.lang.forth
Subject: Re: CASE mis-understanding?
Date: Sun, 09 Feb 2014 22:14:58 -0500
Organization: Aioe.org NNTP Server
Lines: 108
Message-ID: <op.xa1py8wf5zc71u@localhost>
References: <lc33gp$549$1@speranza.aioe.org> <2014Jan27.145730@mips.complang.tuwien.ac.at> <52e7b833$0$24947$e4fe514c@dreader36.news.xs4all.nl> <52e7c363$0$24916$e4fe514c@dreader36.news.xs4all.nl> <lc9dd4$2ob$1@speranza.aioe.org> <66e2b3ae-f7c4-494a-bf1b-0a710763d8f2@googlegroups.com> <0733f401-121c-4072-8563-fd47c0d2a642@googlegroups.com> <u4ydnXyYiIR-VnXPnZ2dnUVZ_gWdnZ2d@supernews.com> <dffbe5ea-9dff-4731-97a3-a7eae1c5299a@googlegroups.com> <op.xakr75ai5zc71u@localhost> <00qdf91cor7adja0stfrn2nrdnom1hq15e@4ax.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@aioe.org
User-Agent: Opera Mail/12.16 (Linux)
X-Notice: Filtered by postfilter v. 0.8.2
Xref: csiph.com comp.lang.forth:28287

On Sun, 09 Feb 2014 00:34:20 -0500, David Thompson  
<dave.thompson2@verizon.net> wrote:
> On Fri, 31 Jan 2014 18:44:19 -0500, "Rod Pemberton"
> <dont_use_email@xnohavenotit.cnm> wrote:
>> On Fri, 31 Jan 2014 00:51:10 -0500, <hughaguilar96@yahoo.com> wrote:

>> > [...] C's SWITCH works correctly.
>>
>> Which one?  C has two: structured and unstructured.  But, I'd say
>> they both work correctly.
>>
> Huh? There's only one switch statement in C, and no other construct
> implements the same semantics.

It seems that the only time you post to c.l.f. is in regards to
C posts but only those from me.  I really don't know why you keep
"attacking" me on C issues ...  Do you know anything about Forth?


C supports two forms of switch().  This is not explicitly spelled
out in the C specifications, but it's _true_ nonetheless.  It's a
byproduct of how the simple language components of C work together,
and work differently, when they're _not_ present.


The first form is known as a structured switch.  It's the form of
switch() that everyone is familiar with.  The switch has a body
encompassed by curly braces:

switch(x)
{
    case ... :
            break;
    case ... :
            break;
    default:
            break;
}

This form of switch() can be nested with other switch()'s with bodies.
It's called a structured switch because it isn't generally coded to
jump into other control-flow constructs, although it can be.

switch(x)
{
    case ... :
            break;
    case 0x10 :
            switch(x)  /* nested */
            {
                case 0x10 :  /* no conflict */
                     break;
                case ... :
                     break;
                default:
                     break;
            }
            break;
    default:
            break;
}

That is nestable, structured switches.


The other form is the unstructured switch.  It has no defined body.
Unlike the structured switch(), it isn't nestable.  This is because
it has no body.  Without a body for each switch, identical case labels
would create a naming conflict.  It's unstructured because it "appears"
that it can jump almost anywhere due to placement of the case labels.
This gives it the appearance of unstructured "spaghetti" code.  There
is no body to break out of, so use of "break;" should be erroneous.
(The ability for a switch to jump almost anywhere in C is true for the
structured switch too.  Although, it's just not generally used.  Duff's
device is one such exception, i.e, where used.)

This is the classic example of an unstructured switch().  It's from
Harbison and Steele's "C: A Reference Manual", 3rd Ed. 1991, page 231:

switch(x)
   default:
   if(prime(x))
     case 2: case 3: case 5: case 7:
       process_prime(x);
   else
     case 4: case 6: case 8: case 9: case 10:
       process_composite(x);

That is an unstructured switch.  'x' is never 0 or 1.  The switch
switches to the case labels for values 2 through 10, thereby bypassing
_some_ of the if() logic.  (I.e., 'else' must work correctly for cases
2, 3, 5, and 7, even though a conditional value for if() was never
computed.  It's up to the compiler as to whether 'else' is an absolute
branch or a conditional branch using the conditional value for if().)
The switch switches to the default label for values over 10 where the
if() is executed calling prime() and executing one of the if() sections,
respectively.


A similar type of example is Duff's device.  It uses a structured
switch(), but it switches into the body of another control-flow
statement, a do()-while block.  (I.e., it demonstrates that a
structured switch can jump anywhere the case labels are placed.)

https://groups.google.com/d/msg/net.lang.c/3KFq-67DzdQ/TKl64DiBAGYJ


Rod Pemberton