Thread: [Dev-C++] C++ standard - Nesting classes.

Hi,

In a project I have been working on I have nested some classes within a main
class that provides an overall functionality. I wanted to make these
internal classes inaccessible to anyone using the main class. I had hoped
that putting them in a private or protected area of the main class it would
stop outside access, but this doesn't seem to work in Dev-C++ (MinGW). Does
anyone know if that is standard C++ behaviour? None of my C++ books say
anything about access specifiers and nested classes, and they say pretty
little about nested classes full-stop. It could reasonally be that only data
and methods are affected by access specifiers, but it's hard for me to be
sure.

I am now looking for an alternative way of hiding those classes, although I
have found some interesting options in using an implementation namespace, I
am still interesting in hearing any other ideas?

Thanx, Jason.

You are right. The following code compiles in Dev-C++:

class A
{
public:
enum errors{ MEMORY, DISK, MONITOR};

private:
class E
{
 errors e;

 public:
 E(errors er):e(er) {}
};

};

int main()
{
 A a;
=20
 A::errors e;=20
=20
 A::E x(e); // Mistake, E is private
}


This shouldn't compile, so i consider it a compiler defect. Borland C++
5.5 gives the (expected) error:

Error E2247 temp.cpp 23: 'A::E' is not accessible in function main()
*** 1 errors in Compile ***


Ioannis

* Ioannis Vranos
* Programming pages: http://www.noicys.f2s.com
* Alternative URL: http://run.to/noicys



> -----Original Message-----
> From: dev...@li...=20
> [mailto:dev...@li...] On Behalf=20
> Of Jason Hardman
> Sent: Wednesday, September 26, 2001 4:03 AM
> To: Dev-C++
> Subject: [Dev-C++] C++ standard - Nesting classes.
>=20
>=20
> Hi,
>=20
> In a project I have been working on I have nested some=20
> classes within a main class that provides an overall=20
> functionality. I wanted to make these internal classes=20
> inaccessible to anyone using the main class. I had hoped that=20
> putting them in a private or protected area of the main class=20
> it would stop outside access, but this doesn't seem to work=20
> in Dev-C++ (MinGW). Does anyone know if that is standard C++=20
> behaviour? None of my C++ books say anything about access=20
> specifiers and nested classes, and they say pretty little=20
> about nested classes full-stop. It could reasonally be that=20
> only data and methods are affected by access specifiers, but=20
> it's hard for me to be sure.
>=20
> I am now looking for an alternative way of hiding those=20
> classes, although I have found some interesting options in=20
> using an implementation namespace, I am still interesting in=20
> hearing any other ideas?
>=20
> Thanx, Jason.
>=20
>=20
> _______________________________________________
> Dev-cpp-users mailing list
> Dev...@li...
> https://lists.sourceforge.net/lists/listinfo/dev-cpp-users
>=20

Thanks for that. I don't currently have any other compilers on my system so
hadn't thought to check. I know that the Borland Compiler is probably the
most standard compliant compiler available, so the fact that it didn't work
on it probably meens it is standard.

I guess I'll have to inquire of the MinGW development community to see about
them adding compliance. Thinking of all this I remember hearing about a
website that contains a copy of the C++ standard, does anyone know where it
it? I will be searching for it, but it would be easier if somebody already
knows about it.

Thanks, Jason.

----- Original Message -----
From: "Ioannis Vranos" <no...@ya...>
To: <dev...@li...>
Sent: Wednesday, September 26, 2001 1:11 PM
Subject: RE: [Dev-C++] C++ standard - Nesting classes.


You are right. The following code compiles in Dev-C++:

class A
{
public:
enum errors{ MEMORY, DISK, MONITOR};

private:
class E
{
 errors e;

 public:
 E(errors er):e(er) {}
};

};

int main()
{
 A a;

 A::errors e;

 A::E x(e); // Mistake, E is private
}


This shouldn't compile, so i consider it a compiler defect. Borland C++
5.5 gives the (expected) error:

Error E2247 temp.cpp 23: 'A::E' is not accessible in function main()
*** 1 errors in Compile ***


Ioannis

* Ioannis Vranos
* Programming pages: http://www.noicys.f2s.com
* Alternative URL: http://run.to/noicys



> -----Original Message-----
> From: dev...@li...
> [mailto:dev...@li...] On Behalf
> Of Jason Hardman
> Sent: Wednesday, September 26, 2001 4:03 AM
> To: Dev-C++
> Subject: [Dev-C++] C++ standard - Nesting classes.
>
>
> Hi,
>
> In a project I have been working on I have nested some
> classes within a main class that provides an overall
> functionality. I wanted to make these internal classes
> inaccessible to anyone using the main class. I had hoped that
> putting them in a private or protected area of the main class
> it would stop outside access, but this doesn't seem to work
> in Dev-C++ (MinGW). Does anyone know if that is standard C++
> behaviour? None of my C++ books say anything about access
> specifiers and nested classes, and they say pretty little
> about nested classes full-stop. It could reasonally be that
> only data and methods are affected by access specifiers, but
> it's hard for me to be sure.
>
> I am now looking for an alternative way of hiding those
> classes, although I have found some interesting options in
> using an implementation namespace, I am still interesting in
> hearing any other ideas?
>
> Thanx, Jason.
>
>
> _______________________________________________
> Dev-cpp-users mailing list
> Dev...@li...
> https://lists.sourceforge.net/lists/listinfo/dev-cpp-users
>


_______________________________________________
Dev-cpp-users mailing list
Dev...@li...
https://lists.sourceforge.net/lists/listinfo/dev-cpp-users

Stroustrup has a link at his pages for a pdf of a very recent draft for
the standard.


Ioannis

* Ioannis Vranos
* Programming pages: http://www.noicys.f2s.com
* Alternative URL: http://run.to/noicys


> -----Original Message-----
> From: dev...@li...=20
> [mailto:dev...@li...] On Behalf=20
> Of Jason Hardman
> Sent: Wednesday, September 26, 2001 4:56 PM
> To: dev...@li...
> Subject: Re: [Dev-C++] C++ standard - Nesting classes.
>=20
>=20
> Thanks for that. I don't currently have any other compilers=20
> on my system so hadn't thought to check. I know that the=20
> Borland Compiler is probably the most standard compliant=20
> compiler available, so the fact that it didn't work on it=20
> probably meens it is standard.
>=20
> I guess I'll have to inquire of the MinGW development=20
> community to see about them adding compliance. Thinking of=20
> all this I remember hearing about a website that contains a=20
> copy of the C++ standard, does anyone know where it it? I=20
> will be searching for it, but it would be easier if somebody=20
> already knows about it.
>=20
> Thanks, Jason.
>=20
> ----- Original Message -----
> From: "Ioannis Vranos" <no...@ya...>
> To: <dev...@li...>
> Sent: Wednesday, September 26, 2001 1:11 PM
> Subject: RE: [Dev-C++] C++ standard - Nesting classes.
>=20
>=20
> You are right. The following code compiles in Dev-C++:
>=20
> class A
> {
> public:
> enum errors{ MEMORY, DISK, MONITOR};
>=20
> private:
> class E
> {
>  errors e;
>=20
>  public:
>  E(errors er):e(er) {}
> };
>=20
> };
>=20
> int main()
> {
>  A a;
>=20
>  A::errors e;
>=20
>  A::E x(e); // Mistake, E is private
> }
>=20
>=20
> This shouldn't compile, so i consider it a compiler defect.=20
> Borland C++ 5.5 gives the (expected) error:
>=20
> Error E2247 temp.cpp 23: 'A::E' is not accessible in function main()
> *** 1 errors in Compile ***
>=20
>=20
> Ioannis
>=20
> * Ioannis Vranos
> * Programming pages: http://www.noicys.f2s.com
> * Alternative URL: http://run.to/noicys
>=20
>=20
>=20
> > -----Original Message-----
> > From: dev...@li...
> > [mailto:dev...@li...] On=20
> Behalf Of Jason=20
> > Hardman
> > Sent: Wednesday, September 26, 2001 4:03 AM
> > To: Dev-C++
> > Subject: [Dev-C++] C++ standard - Nesting classes.
> >
> >
> > Hi,
> >
> > In a project I have been working on I have nested some=20
> classes within=20
> > a main class that provides an overall functionality. I=20
> wanted to make=20
> > these internal classes inaccessible to anyone using the=20
> main class. I=20
> > had hoped that putting them in a private or protected area=20
> of the main=20
> > class it would stop outside access, but this doesn't seem to work
> > in Dev-C++ (MinGW). Does anyone know if that is standard C++
> > behaviour? None of my C++ books say anything about access
> > specifiers and nested classes, and they say pretty little
> > about nested classes full-stop. It could reasonally be that
> > only data and methods are affected by access specifiers, but
> > it's hard for me to be sure.
> >
> > I am now looking for an alternative way of hiding those classes,=20
> > although I have found some interesting options in using an=20
> > implementation namespace, I am still interesting in hearing=20
> any other=20
> > ideas?
> >
> > Thanx, Jason.
> >
> >
> > _______________________________________________
> > Dev-cpp-users mailing list Dev...@li...
> > https://lists.sourceforge.net/lists/listinfo/dev-cpp-users
> >
>=20
>=20
> _______________________________________________
> Dev-cpp-users mailing list
> Dev...@li...
> https://lists.sourceforge.net/lists/listinfo/dev-cpp-users
>=20
>=20
> _______________________________________________
> Dev-cpp-users mailing list
> Dev...@li...
> https://lists.sourceforge.net/lists/listinfo/dev-cpp-users
>=20