Trackback

no comment until now

  1. Mark,

    Regarding Atom, agreed. Left off for the sake of expediency. I would add, though, that something official is needed. If REST can describe _any_ RESTful system, then it would be handy to _formally_ describe RESTful services over HTTP to remove some of the ambiguity. Whether that be incorporating Atom and APP, standardizing on a description language, or what not. That is, REST is the model, and “REST Services” are the implementation.

  2. I guess we disagree there. Roy’s commented before that formal descriptions of REST are problematic. I don’t recall the reasons but I would expect complexity of the descriptions – as well as of the required input in order to validate – would be a large barrier to its use even if you could build a complete model. Description languages are also unnecessary because RESTful messages are self-descriptive.

    IME, there’s exceedingly little ambiguity in the description of REST in Roy’s dissertation. On a few occasions I thought there was some, only to later learn from Roy that it’s covered. IMO, it’s *far* more rigorous a description than most people even need today.

Add your comment now