Re: Fielding on MOVE & COPY

Thanks, Phill.  I understand better what you are after now.

I'd like to leave aside the particular example of annotations, since we
might have a long argument about whether they are metadata or related
documents.  But I agree that it's up to the metadata system to set policy
for whether metadata gets copied / moved with a document or not, whether it
gets updated or not, or whether the user gets to decide.  It implements a
set of rules about MOVE Invariance and COPY Invariance.  

My background is in the kinds of document management systems Steve Carter
listed in his note.  The typical policy there is the one I described, where
a MOVE keeps the document associated with all the same metadata it had
before (well, except for its location if that is an attribute); a COPY
results in a new unique identifier for the new copy, a new value for
creator, a new value for create date, and perhaps for other
system-controlled attributes, and the opportunity for the user to set new
values for user-controlled attributes.  The problem for a metadata system
with this policy is that it has to know when it gets a COPY request whether
that COPY is really going to turn out to be a MOVE, in which case it should
treat the attributes in a different way.  So it seems that you have to be
able to communicate this to the metadata system somehow if you are going to
allow the metadata system to have this policy.

It's certainly possible for document management systems to behave in other
ways, for example, letting a single set of metadata be associated with many
copies of a document; or letting multiple sets of metadata (appropriate for
different users, perhaps) be associated with the same document; or letting
metadata be set for a document that is not online at all.

--Judy
Name:		Judith A. Slein
E-Mail:		slein@wrc.xerox.com
Phone:  	8*222-5169
Fax:		(716) 265-7133
MailStop:	128-29E

Received on Monday, 9 September 1996 18:39:11 UTC