When you read through all of this, at least you know again why you'd rather do XML-RPC. Yes, it doesn't have a really clean formal definition and the original leading implementation from Userland is buggy as hell. But at least there aren't a bunch of academics without real-world experience hanging around in the XML-RPC community who then think they have to indulge themselves in obscure and twisted standards documents...
At Der Schockwellenreiter you can find the original article.