W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > April to June 1997

Re: New properties (metadata) draft

From: Terry Allen <tallen@sonic.net>
Date: Mon, 16 Jun 1997 17:59:40 -0700
Message-Id: <199706170059.RAA30753@bolt.sonic.net>
To: ejw@rome.ICS.UCI.EDU, w3c-dist-auth@w3.org
Jim Whitehead writes:
| Enclosed is the text of a new draft which satisfies the requirements
| for attributes (Section 5.1 of the requirements document), linking
| (Section 5.2 of the requirements), and retrieval of unprocessed source
| (Requirements Section 5.5).  In this draft, an instance of a piece of
| metadata is known as a property.  This draft differs from the previous
| draft on metadata operations (the GETMETA, ADDMETA, DELMETA draft) in
| that it uses GET to retrieve properties, and employs new PATCH method
| to add and delete property values.  A SEARCH method provides the
| capability to search properties.

This is interesting but needs some XML work.  On quick perusal:

| This draft also makes use of the Extended Markup Language (XML) as an

Extensible, not Extended.

| on-the-wire representation for property names and values.  Due to the
| recent release of the Meta Content Framework (MCF) draft, and the
| similarity between MCF and the XML representation given in this draft,
| there is currently debate within the Design Team over whether the
| on-the-wire representation should use MCF.  The primary affect of this
| debate is limited to Section 2.7, Visualizing a Property Tree Using
| XML, which gives an algorithm for converting properties to an XML
| representation, although it would also affect the examples throughout.

| 2.7 Visualizing a Property Tree Using XML
| While it is possible to visualize tree structures in many ways, for use by
| WebDAV, this specification defines a linear visualization of property trees
| using the Extensible Markup Language (XML) [Bray, 1997] suitable for
| transmission as part of an HTTP protocol stream.
| A sketch of an algorithm for creating an XML linearization of a property
| tree is given below.
| Begin by printing an "<XML>" element, if this is a standalone document.

You really want a Document Type Definition for what you're doing here,
even though XML allows documents that don't have a DTD.  In any event,
XML is not a particularly good name for the root element; the XML folks 
are likely to preempt it.  How about WEBDAVTREE, etc.

| Next, perform a traversal of the property tree, collecting the set of all
| schemas used by all properties. Print one "<XML:Namespace>" ... "<AS>" XML

Despite the fact that the MCF spec uses element type names that include
colons, my testing (with James Clark's JADE 0.7) and my reading of the spec
suggest that the colon is not a valid character here.  However, the spec
is fluid, and you should ask for advice on this score, e.g., by posting
to the XML-DEV list.

| entity pair for each schema in use, ensuring that unique tags are used in

I am not sure that "XML entity pair" means anything.  I think you mean
"two start-tags".

| the "<AS>" XML entities. Save the mapping between schema names.
| Next, perform a preorder traversal of the tree. For each node, print its
| name as "<"{unique schema tag}":"{property URI which is relative to the
| schema's URI}">", then print its value, then print "</>". Note that the

</> is not in XML; you need to use a full end-tag, e.g., 
<WEBDAVTREE.Namespace> ... </WEBDAVTREE.Namespace>.  

| process is recursive for values which contain properties. Whitespace may be
| printed between XML entities for prettyprinting.
| Print a closing "</>" (to close the "<XML>") if the document is standalone.

Again, you need the full end-tag.

| For example, the following properties are defined on
| http://somewhere.com/resource.
| <XML>
|  <Namespace><Ref>http://www.w3.org/standards/z39.50/</><AS>Z</></>

should be 


What does the "Z" mean?

|  <Namespace><Ref>AIIM:Dublin:</><AS>D</>
|  <Z:Author><D:PersonalName>Shirley</><D:FamilyName>Mansai</></>
| </>

I see.  You are trying to write a DTD on the fly.  There is no way of
validating this structure, of knowing that you have constructed it
correctly.  Arbitrary extensibility of syntax is nice, but so is
validatability.  Why not construct 

<!doctype webdavtree.mine [
<!element PersonalName (#PCDATA)>
<!attlist PersonalName semantic CDATA "AIIM:Dublin">

?  Then you can parse with a validating XML parser to see that your
element definition structure and document body are correctly constructed.
Further, "AIIM:Dublin" can be a URL or URN.

| To create a reference to the value "Mansai" one would perform the following
| steps.
|   1. Add the DAV parameter to the base URI,
|      http://somewhere.com/resource;DAV.
|   2. Add "/" to refer to the root of the resource's property namespace,
|      http://somewhere.com/resource;DAV/.
|   3. Change the author attribute's name into parameter format by changing
|      the "/" to "!" and encasing it in parenthesis. We must encase it in
|      parenthesis in order to indicate that we have made the "/" to "!"
|      translation. We make the translation in order to prevent confusion over
|      where segments are and to make sure that relative URIs will continue to
|      work.
|      http://somewhere.com/resource;DAV/(http:!!www.w3.org!standards!z39.50!author).
|   4. Next we continue the path by adding the family name attribute. Because
|      it does not include any "/" it does not require encoding.
|      http://somewhere.com/resource;DAV/(http:!!www.w3.org!standards!z39.50!author)/AIIM:Dublin:FamilyName.

If FamilyName is in the AIIM:Dublin spec, why not just assign a normal URL
(or URN) to that semantic, exactly as you do below?  It would require the
cooperation of the spec maintainers, but that would be a real good thing
for interoperability.

| 4 DAV Schema
| The DAV Schema is specified as http://www.ietf.org/standards/dav/. This
| schema is used to indicate support for properties that can be defined on a
| resource and XML entities that can be returned in responses.

If you don't include a version number in that URL you *will* regret it.

| 4.1 Properties
| 4.1.1 Live Property
| Name: http://www.ietf.org/standards/dav/live
| Purpose: To indicate that the parent property has its syntax and semantics
| enforced by the resource on which it is recorded.
| Schema: http://www.ietf.org/standards/dav/
| Parent: Any property
| Values: None
| Description: This property is used exclusively in a response to indicate if
| the resource a property is recorded on is enforcing the syntax and semantics
| of the property. The absence of the Live property tells the client that the
| corresponding property does not have its syntax and semantics enforced by
| the resource on which it is recorded.

Very nice; this is the right approach.  This info could be marked up according 
to an XML DTD too.

| 5.3 Link Property
| Name: http://www.ietf.org/standards/dav/link
| Purpose: To indicate the existence of a link associated with the parent
| property.
| Schema: http://www.ietf.org/standards/dav/
| Parent: Any property that needs to have a link associated with it.
| Values: Src 1*Dst
| Description: Link is used to provide the source and one or more destinations
| of the link. The parent property provides the type. Link is a multivalued
| property so multiple Links may be used together to indicate multiple links
| with the same type.

You might want to synch this up with XML-link, once it's frozen.
(Or again, not, depending on how it turns out.)

| 8.1 Syntax

This really ought to be an XML DTD unless there are technical reasons
it can't be.  XML-lang has the EBNF in it already; there's nothing
to be gained by reinventing it here. 


  Terry Allen    Electronic Publishing Consultant    tallen[at]sonic.net
    Davenport and DocBook:  http://www.ora.com/davenport/index.html
          T.A. at Passage Systems:  terry.allen[at]passage.com 
Received on Monday, 16 June 1997 21:36:31 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:01:11 UTC