Re: [boost] auto_vector - vector with auto_ptr semantics

From: Fernando Cacciola (fcacciola_at_[hidden])
Date: 2002-02-05 15:01:19

----- Original Message -----
From: "Hamish Mackenzie" <boost_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Tuesday, February 05, 2002 4:52 PM
Subject: Re: [boost] auto_vector - vector with auto_ptr semantics

> On Tue, 2002-02-05 at 18:53, Howard Hinnant wrote:
> > On Tuesday, February 5, 2002, at 01:36  PM, Hamish Mackenzie wrote:
> >
> > >> However if the vector uses 1 (move construct) to move the elements
out
> > >> of the way, and an exception is thrown while copy assigning x into
> > >> place, then the vector is still in a valid state.  There are no holes
> > >> of
> > >> uninitialized memory.  Basic exception safety has been maintained.
> > >
> > > But even 1 will leave the source damaged will it not?  So an exception
> > > could still lead to double destruction of the same object.
> >
> > Nope.  Use of #1 to move elements from the end into the unused capacity
> > leaves a vector that is longer than the original with a region of
> > elements in a (for example) default state in the middle.  With #3 these
> > middle elements are destructed.  This half way state must either be
> > valid, or it must be clean-up-able as one starts to copy the argument x
> > in.  I much prefer to keep the vector in a valid state at all times.
> > This approach produces much cleaner, tighter code than doing a try/catch
> > and then putting things back in order in the catch clause.
>
> Ok I see what you mean now. So would this work for option 1?
>
> template< class T >
> void move_construct( T *destination, T &source )
> {
>   using namespace std;
>   construct( destination );
>   move( *destination, source ); // as per previous post
> }
>
> void move_construct( int *destination, int &source )
> {
>   *destination = source;
> }
>
> Also do we need a helper for using this on the stack?
>
> template< class T >
> move_to_here
> {
>   unsigned char buffer_[ sizeof( T ) ];
>   T *get() { return reinterperet_cast< T * >( &buffer_ ); }
>   const T *get() const { return reinterperet_cast< T * >( &buffer_ ); }
>
>[SNIPPED]
>
Unfortunately, this won't be properly aligned for the usage you've shown
below.

Look for discussions about aligned_storage in this list for a summary of the
current state on this issue.

Anyway, if aligned_storage<T>::buffer() were up and running the rest would
be OK.

Fernando Cacciola
Sierra s.r.l.
fcacciola_at_[hidden]
www.gosierra.com