RE: property value clarification

At 12:46 AM -0400 11/2/98, Babich, Alan wrote:
>Albert Lunde wrote:
>"This may be reasonable in some other context; but right now, it looks
>me,
>like WEBDAV intends to treat dead properties as containers which can be
>set
>to arbitrary (internationalized) string values, with no further
>knowledge
>of data type and symantics."
>
>I don't believe that is true. If that were true, then query
>would be hopeless. One can not do query in a satisfactory
>way without facing the data types issue. However, query is
>not hopeless, because your assertion is not true. There are,
>in fact, data types underlying DAV. Otherwise, the model
>would be just junk and should be thrown out. The WebDAV draft
>just does not address data types explicitly. It leaves data
>types to be developed by other DAV efforts, and their
>expression in XML to XML committees.

If we are to require that such schemas be used, then we can design the
protocol ont he assumption that they will be used. If not, we need to be
able to assume something about the properties. I suggest that the choice of
XML (and it's wide adoption by a host of data interchange groups) gives an
obvious choice of data types: that of the language we are using to express
them. The fact that most of the applications of XML that are in the news
right now were _not_ targets of the standard-writing effort are supporting
evidence that this is not a crippling decision.

WebDAV can express that the data represented by a property is the result of
processing that document, in accordance with the schemas associated with
the tags inside the property element.

This allows future use of RDF, or whatever comes down the pike, the use of
plain old DTDs for the conservatives among us, or the use of DTDless
tagsets with informal prose specifications by the adventurous among us. All
these communities are served by knowing what level of data is stored when
an XML property is transmitted.

We probably _do_ want to allow servers to strip comments, for instance. I
suggested James Clark's proposal because he has good judgement, and more
experience of SGML/XML parsing than anyone else I can think of.

The XML data model that scares people so much is strings with labelled
bracketing and attached lists of (property, value) pairs. It can be
_preserved_ by preserving the byte encoding of the data, if processing is
beyond the reach of a server.

It can be sliced, diced, searched and spliced by servers that _do_ parse
the data and _can_ store it.

>"Bringing in the "XML Data" datatypes you refer to would be
>a significant change to WEBDAV."
>
>I don't believe that it is a change to WebDAV at all. Data
>types were there underneath WebDAV all along, so it is not
>a change to WebDAV. Note that data types are being brought
>into *XML* -- not just WebDAV, *all XML* -- by an XML
>committee. XML Data is a change to XML, not WebDAV. Given
>the existence of XML Data and the fact that WebDAV chose XML
>as it serialization format, it doesn't make sense for WebDAV
>to go off and express datatypes some other way. They
>are simply already there in XML "for free".

Right. But this is possible if DAV servers preserve the semantics of
XML-expressed properties -- not if they allow arbitrary rewriting at the
whim of the server.

well-defined rewriting by the server need not be a problem, though of
course eliminating an XML distinction might compromise the ability to apply
some layered standard that uses that distinction.

>Due to the fast pace of the internet, standards generally
>can not create interdependencies on each other. So, WebDAV
>has no dependency on XML Data. However, certain outcomes
>of related standards efforts are generally anticipated,
>and some effort made to mesh smoothly with them when they
>complete. That is also the reason why only a piece of the
>whole DAV design is in the WebDAV draft. Other pieces are
>in other drafts (e.g., Advanced Collections, Versioning,
>DASL, etc.). That is also partly why the property model
>is not completely explained in the WebDAV draft.

Right, and it's why we shouldn't make cryptic decisions like using XML to
express properties, but not expressing how much of the XML semantics are
part of the "real" WebDAV property model (to be defined at some future
date). Standard layering works best when whole standards are applied, and
not some Frankenstein's patchwork of selected "good bits".

>DAV is a very large design coming out in pieces. We need to
>let it come out in pieces. If we wait for all the pieces,
>it would take several years at best, and more than likely
>it would fail.

right. But the pieces that do come out need to be defined clearly enough
that implementing them today won't cause interoperability disasters
tomorrow.

  -- David
_________________________________________
David Durand              dgd@cs.bu.edu  \  david@dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://www.dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________

Received on Monday, 2 November 1998 10:17:19 UTC