Re: structured documents [draft-hopmann-collection-props-00.txt]

At 06:36 PM 2/25/99 PST, John Stracke wrote:

John this is pretty interesting but I think you're going about it in the
wrong order..  First you need to show why webdav applications need a
representation for structured documents, then you need to list the
requirements, and only then should you discuss how to do it.

You made a start when you wrote "to populate a collection, do a MKCOL with
[a request body that is] a compound document".  but that was the only
example you gave, and frankly, I don't think it's sufficient motivation,
because one can already do this with the existing WebDAV methods.
Contrast this with the case that was made for DASL search. While there was
already a way to do it, that way (exhaustive traversal with client side
testing) is so expensive as to be impractical.  What's the problem with
populating a collection using the methods WebDAV provides now?

Depending on your critique, you might decide that, for example, a better
solution would be to add transactionality to WebDAV, rather than compound
documents.  (When one populates a collection, is it important to be able to
set properties (and access rights?) as part of that creation?)

I have no opinion (yet) about whether structured documents should be
expressed in MIME (presumably using multipart) or in XML, and neither
should anyone else, until such time as some requirements are enumerated.

I would encourage you to organize a BOF to pursue this, and perhaps a
working group.  It's not clear it should happen within this working group.

For what it's worth, there's work to design an interoperable
compound/structured document model and API in the DMA community, but I am
not part of it.  Unfortunately this is an industrial consortium no
accessible to the public, but perhaps there's someone on this list who
knows about it 

PS to all: effective March 12, new email: jrd3@alum.mit.edu

Received on Monday, 8 March 1999 12:12:20 UTC