On Mon, 2004-05-24 at 15:58, Andrew Newton wrote:
On May 24, 2004, at 6:34 PM, Bob Atkinson wrote:
It MUST be controlled by the standard and not become a prerogative of
every programmer wishing to publish new definitions. Registering the
XML definitions as a token published with IANA controls this name
space.
[...]
This is not how XML works. Really, it just isn't.
This is not how XML normally works.
I think what Douglas means to say is that there should be a restriction
over and above that which is allowed by XML. Many protocols using XML
do this.
However, an IANA registry will provide an application with no more
knowledge than an HTTP lookup, which is to say almost none. It is
useful only as a place to find the meaning on a token/tag/ns/whatever
and its meta-data (who registered it, where is the spec defining the
semantics, etc...). There is no protocol police to enforce that only
IANA registered tokens appear in the record.
XML specifies neither semantics nor a tag set, but terseness in XML
markup was of minimal importance. Rather than allowing an ad hoc
approach with verbose headers to introduce required definitions, a token
registered by IANA may substitute for both the header that references
these definitions as well as the definitions themselves. A token
registered with IANA then supplants a sizable header with such
references. Any definitional change would then be made as a change to
this token, normally introduced via an adopted specification with
relevant tokens published through IANA. This takes away the ad hoc
nature of XML which will otherwise produce a profusion of definitions.
I see this as a means to clarify a need for an orderly change to these
definitions. This offers no guarantee unpublished tokens will not
appear, but it will be crystal clear such unpublished tokens would be
outside recognized standards, in the same manner using a network
protocol other than those recognized like TCP would be clearly outside
recognized standards. XML "as is" suggests otherwise.
-Doug
Perhaps we should let Jim get the schema written first and then write
up some examples before we seriously talk about such optimizations.
(And I admit that I'm guilty of doing this too.)
-andy