In fact, this transformation exhibits linear runtime behaviour!
As for the potential problem with your transformation -- it is obvious
from the results.
Thank you once again. I am glad that I was right to prefer the more simple
algorithm -- isn't small beautiful? :o)
Dimitre,
1) please repeat your experiment with other processors;
2) are your 3200 nodes nested within each other? This is the only case
when stack overflow can happen. Otherwise, that is, with proposed nesting
levels upto 7, the stack overflow cannot simply be the case.
3) the algorithm with lookahead for the next group is quadratic. by the length
of the nodes on the same level of hierarchy. It can only be linear if all nodes
are numbered sequentially. Which is definitely not the usecase defined.
Please post your data or better mail it to me off-list. This is getting
off-topic
for the list, I think.
Sincerely,
David Tolpin
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list