CURRENT MEETING REPORT


Minutes of the MIME Content--Type for SGML Documents Working 
Group (mimesgml)

Reported by Glenn Vanderburg, University of Texas at Dallas


Overview of Status

o  The types, multipart/related, and access--cid drafts have all been 
   sent to the RFC editors and are awaiting publication as 
   experimental RFCs.

o  Don Stinchfield has informally circulated a new draft of the 
   exchange proposal.

o  Two different proposals are currently being advanced; one is the 
   "encapsulation" proposal (draft--ietf--mimesgml--encap--02.txt), 
   by Ed Levinson, and the other is the "exchange" proposal (draft--
   ietf--mimesgml--exch--00.txt), by Don Stinchfield.  The exchange 
   proposal was described initially as being well--suited for the Web, 
   but also useful for e--mail.  The encapsulation proposal was 
   explicitly designed for e--mail.  Both proposals are intended by 
   their proponents for the standards track, but no working group 
   consensus currently exists.  John Klensin observed that the IESG 
   would not accept two different, non--interoperable specs as 
   proposed standards unless there was both a clear explanation of 
   different circumstances in which to use the different mechanisms, 
   and working group consensus behind the explanation.  Failing 
   consensus on either a single proposal or such an explanation of use, 
   the only alternative would be to advance both proposals as 
   experimental RFCs.

o  There was some discussion of the impact of having working code for 
   one or the other of the proposals.  Bill Smith indicated that EBT 
   plans to implement the exchange proposal in an http--based 
   system, and that James Clark may be doing an independent 
   implementation.  Mark Joseph of Wollongong said that his 
   company has an implementation of the encapsulation proposal in a 
   mail environment.  John Klensin pointed out two process issues: 
   considering that the group charter specifies e--mail----an http--
   based implementation doesn't count.  And in the case of 
   specifications which involve a client/server relationship, two 
   independent, interoperable implementations are required.


o  The point was raised that multipart/related is an important 
   development in its own right, and should not be held to 
   experimental because of disagreement on some dependent proposals.  
   There was general agreement, and John Klensin pointed out that a 
   big reason for the experimental status of multipart/related is its 
   overlap with work done by other working groups, such as the 
   Content--Disposition header.  He also pointed out that multipart, 
   especially with regard to the handling of unrecognized subtypes, is 
   perceived to be one of the more fragile parts of MIME, so any new 
   multipart subtype which was proposed as a standard would be 
   given intense scrutiny by MIME experts.  In such a case it would seem 
   that operational experience as an experimental specification would 
   be useful.


Discussion of the Working Group Scope

o  There was a proposal to extend the charter to cover Web--based 
   interchange, in addition to e--mail.  There was general consensus 
   that Web interchange is also desired, but that it should not be 
   incorporated into this working group's charter at this time.  John 
   Klensin advanced several arguments that addressing the Web issue 
   at this point would cause process problems and introduce further 
   delay.  John also requested a volunteer to co--chair the Working 
   Group, preferably someone  not associated with either proposal.  No 
   one volunteered.


Discussion of Two Competing Proposals

o  There was a request for a description of the relative advantages of 
   each proposal.  A simple description turned out to be difficult.  
   Discussion centered on requirements for various parts of the system to 
   parse the SGML data, and requirements for rewriting parts of the 
   document for interchange (and the effect of such rewriting on 
   message integrity checks).

o  This discussion ended with very little agreement, except on the point 
   that implementation of the encapsulation proposal may prove difficult 
   with some MIME systems which do not convey arbitrary MIME entity 
   headers to viewers or helper applications.



Milestones

o  Two milestones were agreed upon:

o  Don Stinchfield will formally circulate a new draft of the 
   exchange proposal by the end of the year.

o  By the time of the next meeting in March, the Working Group 
   should reach an agreement on how to advance one or both of the 
   proposals.