[boost] Re: In-place construction for variant

From: David Abrahams (dave_at_[hidden])
Date: 2003-11-01 16:16:28

"E. Gladyshev" <egladysh_at_[hidden]> writes:

> --- David Abrahams <dave_at_[hidden]> wrote:
> [...]
>
>> Can you give a "scientific" definition of what you mean by
>> "scientific?"
>
>
> I am not sure... some links/discussions/papers/books
> that elaborate on how and when object invariants
> can be broken and how to deal with that practically.
>
> I'd be especially interested to see how basic guarantees are useful
> in practical C++ programming.  Perhaps some practical usage cases.

Have you read the analysis section of
http://www.boost.org/more/generic_exception_safety.html?

The guarantees are abstractions which help you to reason about program
behavior hiding irrelevant details and highlighting important ones.

> Are there any special considerations for breaking invariants and
> calling other methods of the broken object internally?

Not usually.

> You can imaging some examples like the following.
> struct type
> {
>   //invariant is x + y = 10
>   int x;
>   int y;
>
>   void f( int n )
>   {
>      x = n;  //* invariant is broken
>      f1();  //* is this allowed?

Depends on your coding standard, I guess.  It's not inherently unsafe.

>      y = 10-n;  //* invariant restored
>   }
>
>   //can f1()/f2() or any internal method 
>   //assume that invariant is Ok

Not if called in that way by f; it's up to the class designer.

>   void f1() {}
>   void f2() {}
> };
>
> This becomes much more complicated if you add
> complex polymorphic hierarchies.
>
> It seems that I am not alone who is confused
> about this stuff.

I think you should check out discussions of invariants in literature,
e.g. Bertrand Meyer's discussions of Eiffel.  Then come back and
think of exception-safety in that context.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com