[
Lists Home |
Date Index |
Thread Index
]
If it is an errata change, then it would not
break DTDs. It is a system design change.
XML DTDs, such as they are, are in XML and
are in wide use.
Namespaces are not core and should not be
in the XML infoset unless the infoset itself
is layered into component profiles. That
was an ambitious mistake.
Namespaces aren't a content type.
They are a flag for Web XML processors. One
can write perfectly valid and perfectly
well-formed XML and never declare a namespace.
Different things are going on in the real world.
len
-----Original Message-----
From: Ronald Bourret [mailto:rpbourret@rpbourret.com]
"Bullard, Claude L (Len)" wrote:
>
> What is the problem with keeping them in a separate
> spec?
Very little. It allows Names to have colons in them and still be legal,
but that appears to be the only technical consequence.
> What is the advantage of rolling them in?
Clarity -- the same as rolling errata into a spec. It also formalizes
what's going on in the real world anyway.
> Namespaces are a system feature. They shouldn't
> be in the core any more than XML Schema support
> shouldn't be in the core and XSLT shouldn't be
> in the core. They are framework components.
I disagree. Namespaces are part of the info set -- XML Schemas and XSLT
are not. Furthermore, namespaces are the only part of the infoset that
are not defined in the XML 1.0 spec. Put another way, namespaces are
part of XML 1.0 in a de facto sense, even if they lack de jure status.
> What needs to be fixed as Mike points out are
> the other specifications that all treat the
> infoset differently. That is the mess.
Agreed. The other part of the mess is to get DTDs and namespaces to play
together.
|
