From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,4873305131bf4d94
X-Google-Attributes: gid103376,public
X-Google-Thread: 1014db,4873305131bf4d94
X-Google-Attributes: gid1014db,public
From: emcmanus@gr.opengroup.org (�amonn McManus)
Subject: Mixing declarations and statements (Re: Porting (was ADA and Pascal
 etc))
Date: 1997/11/04
Message-ID: <swutch@kaa.gr.osf.org>#1/1
X-Deja-AN: 287014371
References: <34557f2b.1934172@news.mindspring.com>
 <345AB871.413A@dynamite.com.au> <63d5l4$tub$1@helios.crest.nt.com>
 <878495810snz@genesis.demon.co.uk>
Organization: Open Group Research Institute, Grenoble, France
Newsgroups: comp.lang.ada,comp.lang.c
Date: 1997-11-04T00:00:00+00:00
List-Id: <comp.lang.ada>


fred@genesis.demon.co.uk writes:
>            kaz@helios.crest.nt.com "Kaz Kylheku" writes:
> >What's worse is that mixed declarations and statements are likely going to be
> >part of the new C language.
> Actually that finally resolves one of the biggest bugbears I have when
> writing C. My experience has shown clearly that localising declarations
> can be a big win for readability. Currently I have to use extra { }'s to do
> that in some cases (notably for switch cases). With C9X I will finally be
> able to write code in a natural way according to my preferred coding style.

I think switch cases are a particularly bad example of where this
feature is advantageous.  Where in current C you write either:

    int x;

    switch (thing) {
    case one:
	x = whatever();
	use(x);
	break;
    case two:
	x = maybe();
	use(x);
	break;
    }

or

    switch (thing) {
    case one:
	{
	    int x = whatever();
	    use(x);
	    break;
	}
    case two:
	{
	    int x = maybe();
	    use(x);
	    break;
	}
    }

with the exciting new feature you can write:

    switch (thing) {
    case one:
	int x = whatever();
	use(x);
	break;
    case two:
	int x = maybe();
	use(x);
	break;
    }

Except I very much hope that you can't, since that would be a multiple
declaration of x.  Which x would be referred to in common code that
you arrived at via goto?  Presumably the new C will allow the first
declaration but require the `int' to be left off the second, turning
it back into a lowly statement.  That's terrible programming practice of
course because you're having the meaning of one switch case depend on
the way another one was written.

If you start introducing declarations into your switch cases you're
going to run into this sort of problem all the time.  I did in Java
before I stopped doing it; in fact I find that in Java declarations tend
to migrate up towards the start of the method, where they would be in C.
Nonetheless the new C feature is easy to implement and understand and is
fairly often useful.

,
Eamonn					http://www.gr.opengroup.org/~emcmanus
  "Then I realized I was in France and, in fact, the gibberish was French"
  -- R Johnson