W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > July to September 1998

Re: Hierarchical URLs and Collections (was: Docushare and WebDAV model)

From: <dlaliberte@gte.com>
Date: Mon, 17 Aug 1998 14:19:21 -0400 (EDT)
To: John Turner <johnt@cgocable.net>
Cc: w3c-dist-auth@w3.org
Message-ID: <13784.26191.492072.193300@espion>
John Turner writes:
 > First off, I don't beleve that WebDAV can't have anything to say about the
 > choice of the namespace shape.  It can (and does) require that URL's ending
 > in "/" act like collections 

I didnt think that was true, because it imposes on existing web servers
that mere directories would suddenly become collections.  However, I
think that should be OK, if being a collection doesn't mean servers are
required to do any more than they already do to support the collection.
Still, it becomes kind of meaningless.

 > but can't say what collections exist and what
 > goes in them.  If a site chooses to put all of their documents into one
 > collection and all of their "folders" into another, and not allow any to be
 > created anywhere else that is a perfecly valid choice.  Unfortunately it is
 > not very helpful for someone trying to get at the site with a generic WebDAV
 > client.  Requiring that there also be a hierarchical representation is not
 > something that WebDAV can do.  On the other hand, users of a system probably
 > will :-)

I think there may be a confusion between hierarchical identifiers and
collection containment.  I am distinguishing them, such that the
components of a collection may or may not be accessible via hierarchical
URLs.  If there were never any hierarchical URLs, that doesn't mean
that resources cannot be members of collections; it would only mean
you cannot access the components via extensions of the collection URLs.

Hierarchial URLs do provide a convenient, more direct access to
components of collections, but that is by no means the only way to
access components of collections.  Another way is to first access the
collection and get lists of component URLs.  A particular server might
want to require the latter approach, perhaps to avoid any dependence
that hierarchical URLs impose on the name space, i.e. the resources must
be persistently contained within particular collections forever if the
hierarchical URL is to be persistent.  (There is more to that story if
you want to hear it.)

WebDAV already supports the notion of references to components of
collections, so that is not the question.  Use of references certainly
means that collection containment does not necessarily correspond to
physical containment, but that is true even without references.  But
furthermore, the existence of references raises other concerns such as
containment in multiple collections, and circular references, as John
mentioned.

But regardless of references, we still have a separate question, which
is the original question concerning the use of only non-hierarchical
URLs with hierarchical collections.  I still have not seen any reason
given for prohibiting that combination.  It is not sufficient to just
say it is merely desirable - there must be some functionality that
actually requires it.

--
Daniel LaLiberte
 dlaliberte@gte.com  (was: liberte@ncsa.uiuc.edu)
 liberte@hypernews.org
Received on Monday, 17 August 1998 14:19:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:43:47 GMT