On Oct 1, 2006, at 4:14 AM, Nic James Ferrier wrote:
> Jan Algermissen <jalgermissen@...> writes:
>
> > Hmm...no, Mark is completely right about that. Message self-
> > descriptiveness is an
> > architectural constraint of REST and thus, if your messages are not
> > self-descriptive
> > (aka standardized all the way through) your system does not comply
> > with REST.
>
> I strongly disagree with this assertion. I don't have a link to Mark's
> article where he makes a claim that "self-descriptive" means
> "standardized" so I can't really refute it. But I do refute the
> assertion.

Well, so do I. Self-descriptive means that the type is registered and
the registry points to a specification and the specification explains
how to process the data according to its intent. The specification does
not need to be a standard (a.k.a., a measure that everyone agrees to).
It would help, but most useful standards are defined through use.
Whoever starts sending the data first should define the specification
according to what is being sent, not try to get everyone to agree first.

My question about the API was one of practicality. Given a space that
has been extensively explored, why do people choose to abandon the
available standards? And the answer was the one I was expecting: the
standards are too complex and it seemed better to start with something
simple that is a better fit for the immediate need. There is absolutely
nothing wrong with that choice when it is made with eyes wide open.
You never know what will become the next HTML.

....Roy