On Sep 12, 2013, at 6:23 PM, Howard Hinnant <[email protected]> wrote:
> Need to decrement __data prior to the destruction instead of after:
>
> template <class _Tp>
> inline _LIBCPP_INLINE_VISIBILITY
> dynarray<_Tp>::~dynarray()
> {
> value_type *__data = data () + __size_;
> for ( size_t i = 0; i < __size_; ++i )
> {
> --__data;
> __data->value_type::~value_type();
> }
> __deallocate ( __base_ );
> }
Yesterday, I changed this to:
+template <class _Tp>
+inline _LIBCPP_INLINE_VISIBILITY
+dynarray<_Tp>::~dynarray()
+{
+ value_type *__data = data () + __size_;
+ for ( size_t i = 0; i < __size_; ++i )
+ (--__data)->value_type::~value_type();
+ __deallocate ( __base_ );
+}
Richard wrote:
> Hmm, what should size() return if it's called during the destruction of the
> dynarray?
Once we've entered the destructor, the object has ceased to be. It is an
ex-object.
> During construction, it returns the number of fully-constructed elements.
> Maybe we should do the same here:
That's done because it makes the cleanup easier, rather than worrying about
people calling methods on the object during the constructor.
-- Marshall
Marshall Clow Idio Software <mailto:[email protected]>
A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly
moderated down to (-1, Flamebait).
-- Yu Suzuki
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
