I was very carefull at this part and made lots of tests... You can
try with your local tests, shouldn't be worse than before:) I am
going to update the wiki page first.
--zou
On Dec 3, 2007 3:33 PM, strk <[EMAIL PROTECTED]> wrote:
> On Mon, Dec 03, 2007 at 07:06:15AM +0000, Zou Lunkai wrote:
>
> > +void
> > +DisplayList::mergeDisplayList(DisplayList & newList)
> > +{
> > + testInvariant();
> > +
> > + iterator itOld = beginNonRemoved(_charsByDepth);
> > + iterator itNew = beginNonRemoved(newList._charsByDepth);
> > +
> > + iterator itOldEnd = staticZoneEnd(_charsByDepth);
> > + iterator itNewEnd = staticZoneEnd(newList._charsByDepth);
> > +
> > + while( itOld != itOldEnd )
> > + {
> ...
> > + while( itNew != itNewEnd )
> > + {
> ...
> > + _charsByDepth.erase(itOldBack);
>
> I'm not sure it is safe to use itOldBack again after this erase
> (it' used as the while condition)
>
> > + // replace the old character with the
> > character in the new depth
> > + _charsByDepth.insert(itOldBack,
> > *itNewBack);
> > + _charsByDepth.erase(itOldBack);
>
> Same here, both insert and erase would make itOldBack not point to the end
> anymore.
>
> > +
> > newList._charsByDepth.erase(itNewBack);
>
> This would apply to itNewBack...
>
> > + _charsByDepth.insert(itOldBack, *itNewBack );
>
> and here.
>
> > + // unload remaining characters in old list
> > + while( itOld != itOldEnd )
> > + {
>
> What would itOldEnd point to after all possible modifications above ?
>
> > + // add remaining characters in new list to the old list
> > + if( itNew != itNewEnd )
> > + {
>
> Same here...
>
> --strk;
>
_______________________________________________
Gnash-commit mailing list
Gnash-commit@gnu.org
http://lists.gnu.org/mailman/listinfo/gnash-commit
