Glen Daniels wrote: > > * One feature of the SOAP section 5 encoding is "multi-ref > accessors", which work like this:
It is worth noting that section 5 is optional (though likely high used). Within that option, there are a number of people who consider multi-ref accessors to be a low use case (significantly less than 20% of the traffic). The way a number of pipelined architectures work on the hardware side of the house is to make rather optimistic assumptions along the lines of the high use case and only fall back (by flushing caches and other comparatively expensive operations) to handle the non-common cases. > OK, so as I said, I agree with Sam's points here. The first thing I'd like > to do is some basic performance testing of various XML parsing models. Again, I would like to start even more basic than that...I'll start sketching out a program which simply streams back a response to every request without the overhead of servlets or a parser. Then I'll move up the food chain and be able to quantify the costs of each step. By keeping it simple, I can do equivalent implementations for a standalone program, servlet, DOM, SAX, JDOM, etc. I'll keep everyone posted. - Sam Ruby
