Andre Poenitz wrote:
> Hm. But it would be sufficient for the "users" to use the boost::weak_ptr,
> there is no real need here to hand it out as such?
That is correct. We could have
class InsetBase {
public:
BufferView * cachedview() { return view_.get(); }
private:
boost::weak_ptr<BufferView> view_;
};
> > BufferView * cachedbv = view_.get();
> > if (cachedbv == 0)
> > // do nothing. the parent pointer has been destroyed.
> I see. That's the basic difference to a raw pointer...
As with much boost stuff. Clever, but often overkill. (Is Lars a long way
away?)
> Ok. Maybe we could use the weak_pointer to be on the safe side and have a
> look afterwards whether it is really needed.
Sure.
To paraphrase Pascal (I believe): I am sorry this letter is so long; I did
not have time to write a shorter one.
Often my code is like that :-(
--
Angus
