On Mon, Dec 13, 2004 at 02:26:40PM -0700, Victor Mote wrote:
> Simon Pepping wrote:
> > The code you presented seems to be an algorithm implementing 
> > an iterator over a tree. Because it maintains its state, it 
> > can be stopped and resumed at will, provided you keep a 
> > reference to it.
> >
> > If LMIter would have a reference to its parent LMIter, and 
> > could return to it after having processed all the siblings, 
> > it would realize such a construct. One could stop the 
> > iteration, retain a reference to the active LMIter, and resume later.
> > 
> > Not being dependent on the Java stack would make the 
> > programming much more robust. One would have more freedom to 
> > insert actions, without the fear to lose the iteration state 
> > if one would not carefully return to the same function.
> > 
> > Such a construct would be equally suitable to pull parsing. 
> > The LMIter call to the LM method preLoadNext would request 
> > more child fo nodes, which the parser would provide on this demand.
> > 
> > Do you want to traverse the FO tree, without relying on the 
> > Java stack, and drive the layout process from there by firing 
> > node events?
> Hi Simon:
> You responded to my last posting in this thread, but your questions seem to
> be directed to Finn, so I will let him answer them. Please let me know if I
> have misunderstood.


You are right. I replied to the last message in the thread, but it
contains questions to Finn. Sorry for being unclear.

Regards, Simon

Simon Pepping
home page: http://www.leverkruid.nl

Reply via email to