RE: SD5 - Namespaces - New Version 2

Thank you (Arjun) for your questions.  I will try to clarify what I
meant.

	1.	When I said "The identity of a namespace is independent
of the document (if any) defining its terms" what I had in mind was that
our problem is simply to avoid name conflicts. That is, we simply need
the ability to distinguish names from one namespace versus names from
another.  Having a means of actually knowing what the names mean is a
much more complicated problem, and in some cases not important.  For
example, a firewall program might need to only deal with the meaning of
names from a digital signature namespace. It never attempts to process
any other elements, attributes, etc. (beyond computing a checksum of
their representation). To the firewall, the doctype of most of the
document is unimportant. The firewall only needs the ability to
recognize its elements. Since a document might contain names matching
some names chosen for use in digital signatures, we need to keep the two
spaces separate.  At a minimum, we merely need to distinguish
namespaces.
	     I suspect that all namespaces could have DTDs created for
them, even if only mechanically (as I think was pointed out here
recently). However, for some of the same reasons that XML defines a
well-formed document independent of a DTD, there will be many useful
namespaces that do not immediately have DTDs, or do not have them
readily accessible. (Or, as with a firewall program, the meaning of
various names will be wired into the program.)

	2. "Can I use an attribute from one namespace to qualify a GI
from another namespace?"

	I see the namespace as qualifying the GI, so I don't understand
what it means to "use an attribute . . . to qualify a GI."  This must be
a misunderstanding on one of our parts.

	3. "If I can't, then does the namespace of the GI fix the
namespace for its attributes? Could you clarify this, please?"

	I'll try. I had in mind that it would be tedious and
incompatible with existing documents if we required that every name
carry an explicit qualification.  So, I mean that each element fixes the
default namespace for its attributes and subelements.  This says nothing
about whether various attributes or subelements are valid, merely that,
within a tag, the namespace of the GI is to be used as the namespace for
unqualified attributes, and within an element, the namespace of the GI
is to be used as the namespace for unqualified sub-elements.

	4.	". . . what happens with colons in attribute names? Will
there be an extra well-formedness/validation constraint involved?" 

	I do intend that attributes come from some namespace, and that
means that they, too, will need qualification.  So, yes, colons would be
needed in attribute names. I don't see how this affects the issue of
well-formedness. (Maybe I am missing some insight.)  I see tacking how
we express validation constraints as something we do after July 1.


--Andrew Layman
   AndrewL@microsoft.com

Received on Friday, 23 May 1997 14:58:03 UTC