Minor edits from my review of WD11 - more comments to come.
I'm omitting comments where I've definitely brought them up to Andy Banks in a group review in Hursley.
Line 384: What we describe here is later referred to as a non-shared subscription. We could introduce a definition of subscription which is either a shared or non-shared subscription.
592 & 670: I think this is implying the Server AND client must do UTF-8 checking for all strings. Seems onerous to clients. Is that intended?
885: Implies 0x04 (Disconnect with Will) deletes will message (I did check on TC call that this was an oversight so I'm not raising a separate core issue)
889-892: Implies 0x04 delete will message (i.e. it is same as 0)
898 Says 0x04 deletes will message
983 We discussed this in group review so might already be changed but I think it's important to say session expiry timer starts at end of network connection
1016 Will Message (and any associated will delay timeout) are part of the session state on the server
1674 What if the subscription options are set on a shared subscription. Is that a protocol error?
1827/1854 says server must send properties. Conflicts with 1170 that say if the Request Propblem Information byte is 0 no user properties should be sent.
1888 s/insure/ensure/ (unless we're offering to pay out in the case of a malformed string
1986 Table - I don't understand the difference between 0x80 and 0x83