Brian Smith wrote:
Bill de hOra wrote:
Brian Smith wrote:
That's not what the spec says. What spec text allows a
client to assume a term is globally unique?
The spec says:
The "scheme" attribute [...] identifies a categorization scheme.
We seem to all be unanimous that here "identifies" means "uniquely identifies".
But, immediately prior to that we have:
I'm not sure what thread you've been reading, but in the thread I'm
reading you seem to be the only person here thinking that "identifies"
means "uniquely identifies". The scheme attribute value is an IRI; and
it would make exceedingly little sense for two categorization schemes to
share the same identifier but the RFC does not require or even hint that
the identifier should be unique.
The "term" attribute [...] identifies the category to
which the entry or feed belongs.
Here, apparently, "identifies" does not mean "uniquely identifies" even though
the sentence structure is exactly the same.
The sentence structure is the same because, in both cases, neither
identifier is required to be unique.
I dug around and saw that in draft 04 of the spec., the term attribute was
defined with different language:
The "term" attribute [...] identifies the category
within the categorization scheme to which the entry
or feed belongs.
It is poorly worded but I think it is clearer than what the spec. says now.
That's irrelevant. Draft -04 is not RFC 4287.
I can see that there is consensus here (that I am wrong :), so I
won't argue further about what the spec. says vs. what it intends
to say. If there is some errata mechanism for the specification,
we should consider adding a clarification there.
There's nothing to clarify. The spec is clear. It says exactly what
was intended. There is no errata to document.
- James
See also:
http://www.mail-archive.com/[EMAIL PROTECTED]/msg06874.html
http://tools.ietf.org/html/draft-ietf-atompub-format-04#section-3.7.1
Regards,
Brian
