It could be argued that a JDBC connection and a SQL statement did the same
around database information (i.e. the connection URI) its not quite as clean
as the web approach, but its still a similar concept.

All I'm getting at is that having GUID as the differentiator for REST
probably isn't the best marketing exercise as other approaches (some good,
most bad) have had a similar concept. Because the Web uses DNS to do the
mapping from URI to physical end-point then lots of systems that include a
centralised mapping could also lay claim to the same approach.

One bit of course about the web approach to URIs that means it isn't a
strict GUID is that multiple URIs can refer to the same resource (e.g.
Amazon has a bazillion different URIs that link to a specific book). So it
might be more correct to say that every datasource has _at least one_ URI.

Steve


On 23/07/07, Mark Baker <[EMAIL PROTECTED]> wrote:

  On 7/22/07, Steve Jones <[EMAIL PROTECTED]<jones.steveg%40gmail.com>>
wrote:
> Databases would be an equivalent (but not true GUID), Table + Primary
> Key. But "true" GUID systems tend to be (IMO) pretty dreadful and
> inhouse solutions where every object as a GUID and this is then fired
> around and used as a reference point (often via some horrific central
> server were you pass the GUID and it returns the object.

Your mention of of a central server suggests that identifiers, while
assignable in a decentralized fashion (via GUIDs and their low
probability of collision), couldn't be dereferenced (turned into data)
in the same way. The Web absolutely nailed that by defining a default
mapping between URI schemes and application protocols.

Mark.
--
Mark Baker. Ottawa, Ontario, CANADA. http://www.markbaker.ca
Coactus; Web-inspired integration strategies http://www.coactus.com

Reply via email to