In article <atotic-0801952058390001@nonlinear.mcom.com> (comp.sys.mac.programmer.codewarrior), atotic@mcom.com (Aleksandar Totic) writes:
) Certain kind of class declarations used to work in CW4, and does not in CW5.
) For example:
)
) //class B;      if this line is commented out, compiler complains of "type
) //              mismatch in the program bellow
) class A {
)     class B * b;
) };
)
) class B {
)     short b1;
)     short b2;
) };

Hmm.  This looks kind of like an incomplete declaration of a class B
declared in an inner scope class A to me.  I don't have an ARM handy
at the moment nor do I know whether the ARM is still current on this
topic, but I'd say that if you want to refer to a class B declared in
an outer scope you're out of luck unless you get on the ball and
declare the incomplete type at global scope, or do something like

class A {
    class ::B *b;
};

Well, offhand I'd say that.  However, I just tried out your example
on my copy of CW5 (under 68k) and, well, it compiles fine as is.
My suggestion, OTOH, doesn't compile, yielding a "B is not a
struct/union/class member" error.  Well, ok ....

Perhaps the scope resolution operator can't be applied to class
names in this way; I dunno.  I'm crossposting this to comp.std.c++
for discussion since even if I did have my ARM handy I wouldn't trust
it on this point now after all the tweaks that have transpired
over the past couple of years.  I vaguely recall that only
a class name can follow the "class" keyword, but in that case
I wonder why the class B created in your example above wasn't
treated as a class locally declared in class A.

So, uh, how did you get the example above to produce an error
at compile time, anyway?  As I said, I'm using 68k CW5 Gold and
it works fine for me.

=============== O Fortuna, velut Luna, statu variabilis ===============
uunet!joebloe!joseph   (602) 732-2549 day   joseph%joebloe@uunet.uu.net
1400 N Alma School                #163               Chandler, AZ 85224
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
           Be hip!  Support comp.sys.mac.programmer.moof!

In article <01050166.jcrrqe@joebloe.maple-shade.nj.us>,
joseph@joebloe.maple-shade.nj.us wrote:

> In article <atotic-0801952058390001@nonlinear.mcom.com>
(comp.sys.mac.programmer.codewarrior), atotic@mcom.com (Aleksandar Totic)
writes:
> ) Certain kind of class declarations used to work in CW4, and does not in CW5.
> ) For example:
> )
> ) //class B;      if this line is commented out, compiler complains of "type
> ) //              mismatch in the program bellow
> ) class A {
> )     class B * b;
> ) };
> )
> ) class B {
> )     short b1;
> )     short b2;
> ) };
> ) main()
> )
>
> So, uh, how did you get the example above to produce an error
> at compile time, anyway?  As I said, I'm using 68k CW5 Gold and
> it works fine for me.

Actaully, my example was an abstraction of the problem that occured in my
code, I did not test it:

Here is a real example that does produce an error:

class A {
public:
    class B * b;
};

class B {
public:
    short b1;
    short b2;
};

main() {
    B * b;
    A * a;
    a->b = b;
}

Error   : type mismatch
test.cp line 15   a->b = b;


Aleks
--
Aleksandar Totic
Mozilla Development Team