Lists Home |
Date Index |
> -----Original Message-----
> From: Gavin Thomas Nicol [mailto:firstname.lastname@example.org]
> Sent: 12 February 2002 21:48
> To: email@example.com
> Subject: Re: [xml-dev] REST has too many verbs
> On Tuesday 12 February 2002 06:18 am, Bill de hÓra wrote:
> > The data structure underpinning a HTTP message is a roughly
> a list of
> > key value pairs. The data structure underlying XML is a roughly a
> > tree. As I see it these data structures are appropriate for the
> > respective tasks at hand. What is more solid about trees
> than maps in
> > the HTTP case?
> I think the hey,value pairs have shown themsevles to be insufficient.
> Look at MIME media types for a prime example, or the Accept-*
> Many headers have a built-in ad-hoc syntax for representing structure.
I've run across a similar problem for software agent service
descriptions in a java JSR. There was one key that really wanted its
value to be a sentence in a language. We punted to avoid having to
design a description language that would double as a query language,
hoping to come back to it next time (or hoping someone else would have
done it for us by then ;).
You're right though. The mapped attribute does seem to eventually need
to evolve from a value to a sentence in a grammar that can express
constraints or exceptions for such, as opposed to a value being composed
of a ^ b ^ c or package specialisation. I see this problem when people
start arguing about xml mimetypes for SOAP. But I guess my question is,
is such a language something that should have been built into say, HTTP
with respect to mimetypes, or something that could have been slotted in,
or indeed left to an application to negotiate?
Bill de hÓra