Re: C++ design question
From: Universe (universe_at_tAkEcovad.OuT.net)
Date: 09/28/04
- Next message: Traveler: "Re: Why Can't Software Design Be Like Hardware Design?"
- Previous message: Kurt: "Re: C++ design question"
- In reply to: Universe: "Re: C++ design question"
- Next in thread: Simon Elliott: "Re: C++ design question"
- Reply: Simon Elliott: "Re: C++ design question"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 28 Sep 2004 14:12:10 -0400
Universe <universe@tAkEcovad.OuT.net> wrote:
> "Simon Elliott" <Simon at ctsn.co.uk> wrote:
>
> > I've been advised by the guys over on comp.lang.c++ to post this here
> > to see if anyone can recommend a design improvement. In the following
> > code example, the base class depends on an object in a derived class:
> > fooBase has a reference to an object in fooDerived. It's been pointed
> > out that this is potentially unsafe if bar_base_ref_ is accessed in the
> > fooBase constructor.
> >
> > Can anyone improve on this aspect of the design? Here's the code:
> >
> > class barBase
> > {
> > protected:
> > int i1_;
> > public:
> > barBase(int i1):i1_(i1){}
> > virtual ~barBase(void){};
> > void SetI1(int i1){i1_ = i1;}
> > };
> >
> > class barDerived:public barBase
> > {
> > protected:
> > int i1_;
> > int i2_;
> > public:
> > barDerived(int i1, int i2):i2_(i2),barBase(i1){}
> ^ Please explain the above line. ^
>
> Elliott
The reason I ask is because as I understand doing:
> > barDerived(int i1, int i2):i2_(i2),barBase(i1){}
does not initialize the specific base class instance created as a part
of creating an instance of this derived class if that was the intent.
> > virtual ~barDerived(void){};
> > };
> >
> > class fooBase
> > {
> > private:
> > barBase &bar_base_ref_;
> > private:
> > fooBase(const fooBase &C);
> > fooBase& operator=(const fooBase &C);
> > public:
> > fooBase(barBase &bar_base_ref):bar_base_ref_(bar_base_ref){}
> > virtual ~fooBase(void){};
> > void DoStuffWithBarBase(void)
> > {
> > bar_base_ref_.SetI1(13);
> > }
> > };
> >
> > class fooDerived:public fooBase
> > {
> > private:
> > barDerived my_bar_derived_;
> > public:
> > fooDerived(int i1, int i2):fooBase(my_bar_derived_),
Similarly:
> > fooDerived(int i1, int i2):fooBase(my_bar_derived_),
> > my_bar_derived_(42,43){}
does not initialize the specific base class instance created as a part
of creating an instance of this derived class if that was the intent.
Further there is a "fawlty" attempt to initialize a base class instance
with a 'my_bar_derived' instance where 'my_bar_derived' is only
initialized *after* it is assigned to the base class instance.
> > virtual ~fooDerived(void){};
> > };
> >
> > int main(int argc, char **argv)
> > {
> > fooDerived my_foo_derived(42,1);
> > my_foo_derived.DoStuffWithBarBase();
> > return 0;
> > }
> >
1) The derived classes ctor expressions do not initialize their base
class instance part if that is the intent. They initialize *some* base
class instance, but by the time one gets to initializing a base class
instance the base class instance that is part of the derived class is
*already* constructed and incorporated into the larger derived class
instance.
2) It is definitely not recommended in C++ to attempt to initialize the
base class instance part of a derived class instance from a derived
class instance in the same statement that creates the derived class
instance.
Off the top I would create a structure 'struct' external to both the
base and derived classes that holds the derived classes members that I
want their base classes to interact with.
See "Fragile Base Class' in Struvstrup's FAQ at his web site for more
particulars on the external structure solution.
Elliott
-- dip refers to *abstract interfaces* a form of class abstraction. The opposite form is a concrete implementation class abstraction.
- Next message: Traveler: "Re: Why Can't Software Design Be Like Hardware Design?"
- Previous message: Kurt: "Re: C++ design question"
- In reply to: Universe: "Re: C++ design question"
- Next in thread: Simon Elliott: "Re: C++ design question"
- Reply: Simon Elliott: "Re: C++ design question"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
