RE: CVS runtime updates

[Jean-Yves Baudy <jy...@free.fr>]

Susantha,

It was for my first implementation of simpleTypes (real types 
generated). For the C generation instead of having to force the client 
to alloc anything for those types (has a made type pointers). I created 
the types in the create method. But when deserializing the Array grow 
policy is to double the array. So to free the whole memory a add this 
variable to remender the real size. I think that it should stay.

Jean-Yves

 >Hi Jean-Yves,
 >
 >I had a quick look at the diff and it seems ok except having 
m_RealSize >in
 >Array. Could you explain why you need that ?.
 >
 >Thanks,
 >
 >Susantha.

RE: CVS runtime updates

[Susantha Kumara <su...@opensource.lk>]

Hi Jean-Yves,

> It was for my first implementation of simpleTypes (real types
> generated). For the C generation instead of having to force the client
> to alloc anything for those types (has a made type pointers). I created
> the types in the create method. But when deserializing the Array grow
> policy is to double the array. So to free the whole memory a add this
> variable to remender the real size. I think that it should stay.

Still its not clear to me as to why you need this m_RealSize variable.
 - The Axis engine has the knowladge of the true capacity of the array when
it is deserializing.
 - Its not needed to know the actual capacity of the array anywhere else.
Only the no of available elements (m_Size) is needed.

>
> Jean-Yves
>
>  >Hi Jean-Yves,
>  >
>  >I had a quick look at the diff and it seems ok except having
> m_RealSize >in
>  >Array. Could you explain why you need that ?.
>  >
>  >Thanks,
>  >
>  >Susantha.
>
>