Re: comments on RDF draft 8 Oct 1998: drop Alt

Jim,

Thank you  (again :) for your comments regarding the RDF Model and
Syntax
specification.

Yor raise a very good point...

Without a schema and corresponding application agreements as to
additional properties that clarify the dimension by which the
container members should be seen as alternatives, 'Alt' is indeed
weak.  However, because we expect such schema, application agreements
to be frequently used and documented, and because we expect certain
users to read RDF metadata in XML syntax, we feel that Alt is a useful
encoding hint.  It does not complicate implementations.  If you are
defining a schema and want to do without Alt, you may of course do so
and stick with Seq and Bag.

Additionally, Alt could properly be viewed as only a hint from the
metadata author that in the absence of any other knowledge, the agent
using this data (to display to a user, for example) need choose only
one of the members.  The decision to include this hint is a pragmatic
choice based on expected use of metadata.

Sincerely,

Ralph Swick, W3C Metadata Activity Leader
Eric Miller, Bob Schloss, RDF Model and Syntax Chairs


> Message-Id: <3.0.3.32.19981021133250.009846a0@mailback.parc.xerox.com>

> Date: Wed, 21 Oct 1998 13:32:50 PDT
> To: www-rdf-comments@w3.org
> From: Jim Davis <jdavis@parc.xerox.com>
> Subject: comments on RDF draft 8 Oct 1998: drop Alt
>
> I think you should drop Alt, because I don't believe it can be
reliably
> used.
>
> The point is, Alt conveys an implicit belief by the writer about the
> purpose to which the reader will employ the container.
>
> To illustrate this, consider the two container you employ in example
3.2.1.
>  The first is a Bag of students (in 6.001), the second an Alt of FTP
sites.
>
> Okay, the Alt of FTP sites is implicitly assuming that the reason I
want
> the container is because I want to e.g. download, and downloading from
any
> one site is enough.  But suppose I want the list of sites because I
want to
> upload a new version of xclock?  Likewise the student roster is a Bag,

> which is fine if I want to e.g. ensure that all of them hand in their
> tests.  But suppose I want instead to pick one at random for an
survey?
> Then I'd use an Alt.
>
> I claim that RDF should simply let me state the members of a
container, and
> let the application decide whether for its purposes it needs to visit
all
> the members or any one of them.   The distinction between Bag and Alt
is
> too fragile to deserve being encoded into the underlying
representation.
>
> thank you for this opportunity to comment.

Received on Wednesday, 25 November 1998 14:35:10 UTC