On Mon, Jul 14, 2008 at 07:22:55PM +0200, Abdelrazak Younes wrote: >> We do not if we take the same approach as for MathData and the >> ParagraphsListList: 'new' if needed, 'delete' if it is _really_ >> unused, i.e. _not_ put on the 'other' stack. This means the delete >> can't be in the desctructor as this would be triggered also for >> temporaries (or when the containers would relocate - since they >> are std::deque they won't, but that's more or less a coincidence). > > Well that's exactly what I think I did in my latest patch... But then you > seemed to prefer Richard approach... So I adapted to it...
Hm? Maybe I should stop talking too much then... Andre'
