And now I seem to recall that MS had some condition on use that people could not publish benchmarks of MSXML, which makes that part of my comment unfair. I think some people published a benchmark with a mussing name for that product. But the memory is not strong...
So let me rephrase: in order make to a legimate comparison of a new parse method, it is necessary not just to pick a couple of parsers you consider typical or median or popular. It is better to show the best and worst two plus whatever parser is most often found in other papers: if you have a four-fold improvement over a mediocre but popular parser, and it turns out there are several other parsers that offer a ten-fold improvement, it changes what may be concluded by the reader about your method.
Cheers
Rick
