Re: proposed TAG issues: uniform resource version info and access of resource metadata

On Mon, 25 Nov 2002, Paul Prescod wrote:

> I believe that WebDAV+DeltaV are intended to solve these problems. If
> they are not adequate for some reason, it would be helpful to hear why.
>   Paul Prescod

#1. There should be a uniform way to declare version history of web
resources (recommended by W3C)? ...

I think the idea of WebDAV+DeltaV is very good and hopefully it gets even
more broader attention in W3C than it currently has. To me it seems that
versioning hasn't really been a priority issue (e.g. addressed in the
requirements of XML-schema and XML-sig). Of course, taking everything into
account is next to impossible.

However, I wished that a lightweight revision profile would get
established, providing means for managing version history resources and
version resources manually (without the new HTTP-like transaction
methods). Perhaps achieving this is simply a matter of writing guidelines
how to apply WebDAV+DeltaV protocol with W3C technologies--despite the
fact that some core features would of course be lost? If such guidelines
exist, I'm not aware of them (which of course is only my fault).

However, one thing WebDAV+DeltaV might be "missing", is a simple, explicit
acknowledgement about "substantial changes" between version resources.
Dead properties and baselines (with subbaselines) probably do provide a
syntactical way for doing this but the simple semantics for
major/minor/revision versions (1.1.22 -> 1.2.0 implies "now I must
reconsider this resource") seem a bit blurred (well, to me anyway but then
again, I haven't dug into this nor am I an expert of WebDAV+DeltaV).
Perhaps semantics is a prior assumption.

#2. There should be a "clean", uniform way to refer to (and thus access)
the metadata of web resources?

To me it seems that this is something WebDAV+DeltaV isn't quite addressing
since the WebDAV+DeltaV properties include only(?) those properties
attributed via the transaction model (PROPPATCH), i.e., ignoring e.g. the
RDF statements embedded in XHTML documents, external documents etc. (Or
perhaps WebDAV+DeltaV can indeed be interpreted to encompass all metadata?
[too centralised requirement to be realistic?]) For brevity, I'll leave
the URN part of the discussion aside for now.

What I'm thinking of is simply a uniform "method", a singular access point
for referencing (and retrieving) the RDF statements "declared" by a
resource. In other words, I considering very practical means to simplify
the process of accessing metadata (to give SW/WS applications a boost).

Since RDF aims providing a clean way for making simple statements, it
would seem sensible to provide a uniform way for accessing those
statements as well (and providing a generic structure of pointers to the
actual metadata [those n+1 WSDL, RSS, RDDL, DAML+OIL, DC cards, MIME type,
version information, etc. which to date have no universal declarative
semantics nor naming conventions]).

In addition, transparent approach of working with RDF files (that can be
produced in many ways; also automatically from a WebDAV database) would
seem more versatile approach when compared to expecting all the HTTP
servers to be updated.



Received on Tuesday, 26 November 2002 04:26:12 UTC