Ian Jones, Chair

Doug Bunting
Dale Moberg
Mike Dillon
Matt MacKenzie
Jeff Turpin
Cliff Collins
Jacques Durand
Colleen Evans
Himagiri Mukkamala
Kazunori Iwasa
Brian

Regrets

1. Outstanding issues from previous meeting

Reliable messaging
Ack and bundling
The discussion on the proposal continued from last weeks call. A formal motion was made to accept the proposal to prohibit the bundling of unrelated acknowledgements. The motion was accepted by the call as similr to that described in the minutes of the last call:


1. - The errata document will state that The existing MessageData / RefToMessageId element MUST be identical the Acknowledgment / RefToMessageId element .
2. - The errata document will contain a warning that the Acknowledgment / RefToMessageId element was being deprecated and was scheduled for removal in future releases.
3. - Future versions will not support the feature.

Action Ian to post the above for final comment on the list.


Errata progress
Mike had posted an initial draft of the errata document which everyone accepted the content but the discussion quickly turned to the format and if word should be still be the format. (Ed. Note the PDF versions a generally considered the definitive versions the web site.)

The discussion on the use of DocBok for future versions of the specification was lead by Matt who had posted some details to list which people had looked at. Most people were in favour to move to this format as it had several advantages especially in referencing from future errata and discussion lists. The only concern was the need to convert from the existing word format, this should be addressable if the conversion process is split amongst different people. Matt agreed to post some more details and an example build package to this list.

Action Matt to post a package detailing how to use the DocBook components



2. New Items

Use of filename field in MIME Headers
A discussion on the possible use of the filename field in the MIME content headers was started as the specification makes no explicit mention of them. This may lead to implementation ambiguities. Doug agreed to lead on looking at this and the team will decide if we wish to pursue this further.


Use of ACKS in retries
A discussion on how and when ACKs should be resent when reliable messaging is in operation began which lead to no conclusion. The issue raised was what should the acknowledgement be:


1. - should it acknowledge the original message if it was received i.e. is just the original ack resent or
2. - should it ack the resent message i.e. it is a completely new ack.

1. Ian to post final proposal on ACK bundling
2. MMatt to post a package of DocBook details