PROP_ATTR: consensus points

OK, let me see if I can summarize the points of rough consensus in this
discussion of the PROP_ATTR issue. From reading this thread, there were
several distinct issues raised:

* Should the server persistently store XML attribute information: (a) on the
property contents, and (b) on the property name element.

* Is XML a marshalling format, or MUST the server store property information
as XML?

* What is the behavior of the server on round-tripping XML attribute
information of all kinds (including XML namespaces, xml:lang, and other
standard, and user-defined attributes).

I believe we have come to consensus on the first two issues:

1. The server MUST persistently store XML attribute information stored on
XML elements contained by the XML element whose name is the name of the
property.

(A simpler way to describe this is: The server MUST persistently store XML
attribute information stored on XML elements in a property's value. But,
since we haven't agreed yet on what the "value" of a property is, I wanted
to express the consensus point without using the "value" term).

2. I believe there is *rough* consensus that XML attributes on the property
name element MAY be persistently stored, unless explicitly noted as an
exception in the protocol specification (i.e., XML namespaces and xml:lang.)

3. XML is a marshalling format for the name and value of a property; the
server is free to use whatever persistent storage format it desires.

I was not able to detect consensus on the round-tripping issues.  However,
this is far enough away from the original scope of PROP_ATTR that I have
created a new issue for this, called PROP_ROUNDTRIP.

- Jim

Received on Monday, 16 April 2001 13:54:53 UTC