CURRENT_MEETING_REPORT

CURRENT_MEETING_REPORT_

Reported by Sam Wilson/The University of Edinburgh

Networking Multimedia Applications BOF (MULTIAPP)

Chris Adie introduced himself as the leader of the RARE Multimedia
Information Services Task Force and described the scope of the MULTIAPP
BOF, namely covering networked access to multimedia (MM) resources from
both the user's and provider's points of view. In particular it was not
intended to cover MM conferencing, which was being addressed elsewhere.

Three developments suggested the need for work in this area:

o Users were using MM authoring tools (e.g. Guide, Toolbook) to
produce MM applications;

o Existing client/server tools such as Gopher, WAIS and WWW are being
extended to cope with MM; and

o Various MM-related network tools and projects have emerged: MICE
(in the conferencing field), RTP (Real Time Protocol), ST2 and
multicast (though multicast was currently being used largely in the
conferencing area: person-to-person rather than
machine-to-person).

Chris then gave a short presentation covering the probable application
categories, the requirements, the existing systems and standards and the
aims he saw for the BOF. The slides from his presentation follow these
minutes. A couple of points of interest to go with the slides:

1. Chris is the editor of RARE Technical Report 5 ``A Survey of
Distributed Multimedia: Research, Standards and Products''
available on paper and in various forms for anonymous FTP on
ftp.ed.ac.uk in the directory pub/mmsurvey; and

2. Many projects seem to use SGML in one form or another.

A list of relevant issues that might be suitable areas of work was then
solicited from the floor. The discussion arising from this was wide
ranging and it took some time before suitable areas for IETF involvement
was agreed upon. The list eventually comprised:

o Responsiveness
o Reliability
o Isochronism
o Reverse Control Flow
o Hypermedia Linking
o Presentation

Topics covered and points made in the discussion included:

- The first three items in the list fall into the 'Quality of Service'
area; the third item, 'Reverse Control Flow' might fall into either
QoS or into 'user requirements' since it might refer to low level
control ("I'm a slow display, send me fewer frames per second") or to
the users navigation through the MM application ("send frames 36 to 90
at 1 frame per second").

- In the QoS area questions arose about whether 'responsiveness' (for
instance) was a 'lower layer' problem and should therefore be tackled
by the network providers; it was noted that caching strategies can
help there.

- Also in the QoS area there was a lot of talk about how to handle the
syncing of data, both in the 'isochronicity' area and also in the
realm of synchronising different parts of a MM presentation; RTP
provides time stamping and sequencing but requires higher level stuff
to actually use this information.

- Handling of 'out-of-sync' conditions need to be handled in consistent
ways - Apple's Quicktime seems to do it impressively and so may other
architectures (Microsoft's "ClockWorks" was mentioned).

- What can the IETF contribute to an area where there are already so
many competing products, some being pushed heavily by the big
manufacturers? Some answers to this covered the areas of interworking
and platform independence - the ability to play back and control MM
applications across the network and onto a different platform from
that for which it was written.

- In the area of 'Hypermedia linking' it was pointed out that what is
needed is more than hyperTEXT linking - we need to be able to define,
say, active areas of graphics as well as active words or phrases.

At this point Tim Berners-Lee was invited to give a short description of
HTML and HTML+, the description languages underlying WWW, and compared
it particularly to HyTime. Tim was less than encouraging about HTML+ as
a contender for a generalised MM presentation language or interchange
format, but Chris Adie was more hopeful.

At this point the group seemed to be inclining towards the view that
there was nothing obvious that the IETF could be doing except perhaps to
encourage a pilot project to mount a MM application across a network.
Discussion re-focussed on the description of interchange formats and
`Multimedia X.' Carsten Bormann of Technische Universitaet Berlin
revitalised the meeting by suggesting the following `taxonomy' for the
problem:

o Remote Access to MM Applications
Needs MM analogue of X Window System (note that RTP provides some
of the necessary functionality but higher level integration is
needed along with some kind of session control).

o Global Hyperlinking

- URI/URL
- searching
- ``location address'' (this is a HyTime term)

o MM Document/Application Interchange Format
This is already being tackled by, for example, HyTime, MHEG, and
HyperODA.

The major area of work here is in the ``MM X'' area, though there is
some work needed in interchange formats. It was decided that some
members of the group (which members seemed to be decided informally
later) should:

1. Take one or more existing applications;
2. Try to mount them for access across a network;
3. Decide whether it is currently possible to achieve this; and
4. Come back to the IETF within a year or so to see if there was any
standardisation effort that should be followed up or whether the
existing tools and techniques were sufficient.

Looking further at the X analogy brought Carsten to the overhead
projector again to describe the possible structure of ``MM X'' (or
perhaps just ``M'' a Multimedia X analogue.

"monomedium X" | "MultiMedia X"
------------------------|------------------------
|
Events |
Drawing requests | Video and audio clips
Windows |
|
Window Manager(s) | QoS Mgr, Time Mgr,
| Schedule Mgr, etc.
-------------------------------------------------
|
TCP | RTP
|

Between the two layers in ``MM X'' would be specifications for
synchronisation and playout. Some people (SunSoft, DEC, others that
people vaguely recalled but couldn't name) are already working in this
area.

Discussion followed on the details of what work should be done, how the
tasks should be limited (need achievable goals), and how to liaise with
other groups. The meeting eventually settled on the following items:

o Attempting some pilot work in mounting (at least) one MM
application across the network;

o Working up a draft charter for a proposed IETF working group;

o Looking at ways of linking existing or emerging standards (RTP,
HTML+, etc.) to define a MM access protocol somewhat analogous to
X; and

o Study of prior art and liaison with other groups (SunSoft, DEC,
Apple, Microsoft, Bellcore and various other groups and consortia).

Minute-Taker's Disclaimer: In the tradition of BOFs, the discussion in
this one was varied and wide ranging. Trying to make coherent minutes
that reflect the feel of the meeting is a tricky task and has no doubt
been influenced by what the minute-taker happened to be concentrating on
(or not) at any particular moment. I therefore take full responsibility
for any errors of fact, substance or emphasis, whilst denying any
liability in that regard!

Attendees

Chris Adie [email protected]
Axel Belinfante [email protected]
Carsten Bormann [email protected]
Steve Buchko [email protected]
Jill Foster [email protected]
Kevin Gamiel [email protected]
Joan Gargano [email protected]
Terry Gray [email protected]
Maria Heijne [email protected]
Rune Hjelsvold [email protected]
Xinli Hou [email protected]
Sascha Ignjatovic [email protected]
Xander Jansen [email protected]
Ola Johansson [email protected]
John Johnston [email protected]
Thomas Kaeppner kaeppner%[email protected]
Jim Knowles [email protected]
Steen Linden [email protected]
John Martin [email protected]
Paul Milazzo [email protected]
Ronny Nilsen [email protected]
Lars-Gunnar Olsson [email protected]
Geir Pedersen [email protected]
Dan Romascanu [email protected]
A. Velu Sinha [email protected]
Simon Spero [email protected]
Guido van Rossum [email protected]
Ton Verschuren [email protected]
Chris Weider [email protected]
Sam Wilson [email protected]