Lists Home |
Date Index |
Who's job is made easier by a choice to put namespaces into the base XML spec?
There is nothing about XML as currently defined that restricts it to the Web
platform. This is a good thing because one can use well-formed XML
in ways that never use URIs or can use URIs at their convenience.
The WWW is a platform. If we say, it operates in a disconnected mode (on a
laptop on an airplace), a definition of the Web boundary which includes the notion
"information on a network" is wrong, or the definition of "network" has been broadened to mean
any environment where URIs are used. Resorting to abstractions of "information space" is
metaphorical hokeyness to justify definitional boundary expansion.
If namespace URIs are resolvable by whim, introducing them to the core
definition of the system increases the unreliability of XML systems
operating in a disconnected mode. Note that this is not a flaw
in XML; it is the result of using location identifiers as
resource names in a framework that has to rely on context or
whim to choose the role and apply the processor.
XML Doesn't Care. Namespaces care or don't depending on who one believes
or the system implementation.
As namespaces items are InfoSet items and the InfoSet doesn't equal XML,
then namespaces are provided and the InfoSet rec provides for their use in XML.
It seems to imply that InfoSet is core but that is how things are evolving anyway.
What reasons are there to change this? See first question.