Re: [boost] [non derivable] Submission for Review

From: Manuel Fiorelli (manuel.fiorelli_at_[hidden])
Date: 2007-03-08 14:35:57

I agree with you, when you say that BOOST_NON_DERIVABLE macro is not useful
with structures,
since it introduces a great overhead.
I was obliged to introduce that virtual functions, in order to fix a bug of
the first version, but
maybe I will find an alternative solution. The problem which I refer to is
the following

class Foo : BOOST_NON_DERIVABLE
{

};

class Goo : public Foo, BOOST_NON_DERIVABLE
{

};

If you use the previous version it compiles, while now it will cause a
compilation error (as expected).

2007/3/8, Yuriy Koblents-Mishke <yurakm_at_[hidden]>:
>
> Positive:
>
> The macro lets guarantee that a class is final.
>
> For example, there is a "textbook" error of deleting an object of
> derived class whose base class have non-virtual destructor. The error,
> resulting in splitting, manifests in run time and is hard to diagnose.
> The usual way to avoid the error is to declare almost all classes with
> virtual destructor, even when it is not needed. The macro provides a
> better alternative.
>
> Negative:
>
> An example: a point on a screen. In a graphic application it can be
> represented by a structure of two unsigned int, x and y.
>
> Can we make the structure final?
>
> The class nonderivable_helper contains a virtual function and a
> virtual destructor. Every class derived from nonderivable_helper also
> will contain virtual function and/or destructor. It follows that in
> typical implementation the final classes will contain a pointer to
> vtable, even if they have none virtual functions of their own (or none
> function at all).
>
>
>
> Technical:
>
> I tested a previous version with MinGW gcc 3.4.2 and with MSVC 8; the
> both worked as expected.
>
> Sincerely,
> Yura
>
>
>
> On 3/7/07, Manuel Fiorelli <manuel.fiorelli_at_[hidden]> wrote:
> > I wrote a library which implements the "non derivable" semantics, which
> is
> > natively supported by some programming languages such as Java or C#.
> > I  believe it is mature for a submission and I request everybody for a
> > feedback about the library itself.
> > I could test it only witch GCC 4.1 but it seems to be 100% standard
> > compliant.
> >
> > I published the library in the Vault under the folder "Generic
> Programming"
> > with the name "nonderivable.zip".
> >
> > I know that in the vault there is a similar library (called
> > "noninheritable")  but one must consider that:
> >     1) it uses a different idiom (private constructor and friend
> > declaration)
> >     2) it uses non standard features, while my implementation aims to be
> > 100% standard compliant
> >     3) it uses a different syntax
> >     4) at least my request could increase the interest about that
> library
> >
> > Usage -------/
> > Suppose you want to define a final class Foo, then you have to write
> >
> > class Foo : BOOST_NON_DERIVABLE
> > {
> >
> > };
> >
> > Best regards,
> > Manuel Fiorelli
> > _______________________________________________
> > Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
> >
> _______________________________________________
> Unsubscribe & other changes:
> http://lists.boost.org/mailman/listinfo.cgi/boost
>