Lists Home |
Date Index |
In the beginning...
There were documents and documents had names and all was good. And on the
seventh day God created the hypertext transport protocol and man could get
documents from names. And documents had fragments and fragments had names.
And man prospered.
But man desired knowledge. Man spoke of "concepts". Mankind was divided into
many camps and each use the same words to mean different things. One of the
camps was called "REST" another "RDF" another "TM". Man stopped working and
spent countless hours debating knowledge. And no knowledge occured. And
there was a recession.
One camp extends the meaning of an HTTP URI to identify whatever the creator
intends, specifically an "http" URI need not only identify a _document_
rather an arbitrary resource. Fielding 2000
REST accomplishes this by defining a resource to be the semantics of what
the author intends to identify ...
The concept is that the resource which is described by a document entity is
what the document is _about_, not the document entity itself.
A second camp (RDF) maintains that an HTTP URI identifies a document. In
this case the class of a URI is consrained by the URI scheme: "http" is for
documents, "ftp" is for files, "person" might be to identify people.
To identify concepts 'within' a document, the fragment identifier is
employed. A URI reference is used to identify concepts:
The syntax of a fragment identifier is defined according to the media type
of the resolved entity, fragment identifiers being used by the user agent to
obtain a fragment of a returned entity.
It is not clear that, a URI reference can be assigned semantics outside of
the URI resolution process.
Furthermore, if a URI identifies a document, a fragment identifier would
seem to identify a fragment of the document, not a concept.
1) A URI is considered opaque outside the resolution process.
2) The scheme of a URI does not in and of itself assign semantics to the
resource which is identified..
3) A URI reference is be defined as identifying a "sub resource". The
fragment of a document entity identified by a fragment identifier has
roughly the same relationship to this "sub resource" as does the enclosing
document entity to the enclosing resource. A "sub resource" can thus be
considered a node within a graph.
4) A media type independent fragment identifier syntax should be defined
which would roughly correspond to the _syntax_ of XPointer (e.g. raw names