> But what happens when the user wants to have an xml stream in the > metadata? For example when a bot is using the ESME api and wants to > store additional material in the message? I was trying to duplicate > the Twitter Annotations behavior: > http://apiwiki.twitter.com/Annotations-Overview
Currently when the client gets the metadata as a string it should be unescaped, which should give the expected result parsable both in the JSON or XML case. The correct thing for a client to do is to also URL encode the request when posting, which is also a requirement for the Twitter API, and it will not have to deal with shell escaping issues such as when you are using curl from the command-line.
