MQTT-SN SC Meeting

Timezone for this meeting is BST (UTC+1)

Agenda:

• Roll call

• Approve agenda 
  

• Approve minutes from last meeting - https://groups.oasis-open.org/higherlogic/ws/public/document?document_id=72831&wg_id=3d1c9127-09c1-4b2e-8801-018dc7d3f558

• Voting rights update

• Tasks done:

  • requested template for implementation guide

  • We had a workshop on 28^th May – a summary is here: https://groups.oasis-open.org/higherlogic/ws/public/document?document_id=72834&wg_id=3d1c9127-09c1-4b2e-8801-018dc7d3f558. As a result Ian did the following:

    • Moved ADVERTISE/GWINFO/SEARCHGW to end of section 3

    • Moved C.2 Gateway Advertisement and Discovery to end of Appendix C

    • Added revision history for the work done in the last year

  • Updated PINGREQ actions

  • Updated all links from “section x.x” to “x.x section name”

  • Fixed sleep packet exchange diagram to use SLEEPREQ/SLEEPRESP rather than DISCONNECT

  • Added retry timeout transitions to Client and Server state diagrams

  • Moved Client and Server state diagrams to Appendix C.5

  • Added informative comment to 3.11.3 about issue 649 - Gateway receiving a different Client Id in PINGREQ when sleeping

• Issues/questions:

  • Ian suggests moving the Advertise/GWINFO and SEARCHGW packet types (2.1.3) also to the end. If we also move REGISTER/REGACK, we can use the same first 16 values as MQTT.

  • Simon’s question about CONNECT in the Asleep state: We meed to talk thru if there are semantics that mean CONNECT is changing something that the client may otherwise want

  • Davide suggested removing the Forwarder Encapsulation. Ian’s response: The forwarder sits at the boundary of different network technologies that do not directly communicate with each other, for instance between ZigBee and UDP. So I think there's a better case for this wrapper than advertise and discovery.

  • Andy suggests picking one of the terms Server or Gateway and using it.

  • How optional should waiting for a SLEEPRESP be? (see Davide’s comment in 4.14)

  • Simon’s question in 4.4.2 – don’t we need to increment the packet id on retry in some cases (CONNECT)

  • Section 3.6.2 – where should we provide an example on how to associate a PUBLISH (or any packet) received by a GW with a session handled by the same GW? Implementation Guide?

  • Davide’s comment: references to multicast can be removed if Gateway Discovery is removed. I think we can remove references to unicast/multicast anyway, by rephrasing.

• Working draft todo:

  • review and update entire draft for use of Packet and Application, Will and Retained Message terms

  • gather conformance statements and compete Appendix B

• AOB

• Confirm time and date of next meeting