Simon Fell > Its just code
Simon Fell would rather see a general binary infoset serialization format, that didn't rely on schema's being known a priori. Why? WSDL provides meta data that is not known a priori and it seems to be working OK. The key, IMO, is to allow the meta data to evolve on one end of the wire without requiring it to evolve on the other end of the wire at exactly the same instant. [deem] That's one of the keys. In addition, you'll have to cope with some of the open ended schema constructs anyway (like any & anyAttribute), and tying it to schema just feeling like you're cross connecting to many layers.
What do you think about a binary XML serialization format? How much of the ubiquity of XML do we owe to text and angle brackets? Does the future success of web services depend on them or on getting rid of them?
Has anyone else run across this (also this)? [deem] There's a number of cases where this is useful, given that most people using SOAP don't give a monkey's about the resulting XML, the bits going over the wire could easily be XML with angle brackets, a binary Infoset serialization or whatever. Now having said that, from a development / debugging perspective, having it in plain text makes life much easier. I've been trying to debug some issues with a SSL only service, and its driving me nuts trying to workout what's going on, without being able to see the actual "bits on the wire". Working on the DIME support for PocketSOAP prompted me to add a hex viewer mode to YATT. The articles Mike point to describe what appears to be a schema specific binary encoding, this is an interesting idea, but personally, I'd rather see a general binary infoset serialization format, that didn't rely on schema's being known a priori.
Reactions to the Google API. Paul Prescod has been "working like a fiend" on a plain HTTP alternative to Google's API. He's put up a summary page on the various efforts, including a petition to Google. [More Like This WebLog]