Shabtay’s take 04/28 - Ok with partial acceptance of the message with eom_auto_flush disable.  Recommend that we don’t allow this with eom_auto_flush enable. The term eom_auto_flush is quite of cumbersome.

Committee agrees still active: John agrees and says we need better terms.

Synchronization semantics of FLUSH for non-blocking.

Discussed flushing, particularly nozzles with partial messages. Questioned behavior with messages larger than the buffer size, when used with the non-blocking API. How much user responsibility to size the buffer, and whether the blocking API should be allowed to return a partial acceptance of the message.  John, Per like this idea, Shabtay wants to look at it more.