[osg-submissions] Group::removeChildren fix

[Bradford, Chase]

> 
> Hello Robert:
> 
> Yes,  I forgot the prototype was for position and num elements.  Sorry
> about that.  I guess the confusion is:
> 
>          _children.erase(_children.begin()+pos,_children.begin()
> +endOfRemoveRange);
> 
>          childRemoved(pos,endOfRemoveRange-pos);
> 
> Then we need childRemoving:
> 
>          childRemoving(pos,endOfRemoveRange-pos);
> 
>          _children.erase(_children.begin()+pos,_children.begin()
> +endOfRemoveRange);
> 
>          childRemoved(pos,endOfRemoveRange-pos);
> 
> If we want to have access to the actual objects before whacking them
> from the list.
> 
> 
> Take care
> 
> Garrett
> 
> On Jan 14, 2008, at 2:13 PM, Robert Osfield wrote:
> 
> > HI Garrett,
> >
> > On Jan 14, 2008 5:38 PM, Garrett Potts <potts at cfl.rr.com> wrote:
> >> Is there a problem with childRemoved taking as an argument a
> >> osg::Node* type.  Looking at the code in where childRemoved in the
> >> osg::Group node the indices as mentioned can't really be used to
> >> access the list in which the node was removed.  Instead, would it
be
> >> better to just pass the osg::Node* to childRemoved?
> >
> > The unsigned int position is used because the Nodes themselves have
> > been removed and potentially no longer exist, so it'd be very bad
> > thing to pass the Node* around.
> >
> > The concept behind the childInserted and childRemoved is that it
> > enables the maintenance of data structures that exist on a one per
> > child basis where there is 1:1 correspondance between the lists that
> > hold the original child data and the local data.
> >
> > Robert.
> > 

For your case, would it be sufficient to override removeChildren(
unsigned int pos, unsigned int cnt), and do your processing before
forwarding the call to osg::Group::removeChildren(pos, cnt )

Chase