NEW ISSUE: RMS state table and SequenceClosedFault

[Matthew Lovett <MLOVETT@uk.ibm.com>]

This issue is raised as a follow on from
the F2F, to address action item 0062.

Title: RMS state table and SequenceClosedFault
Description: The RMS state table currently
has many placeholders in Row 16, which describes the RMS state transitions
when it receives a SequenceClosedFault. The author of the table considered
the spec unclear, and so was unable to complete this row.
Justification: The spec should be complete,
or explicitly declare some behaviour to be beyond the spec.
Target: core
Type: design

Proposal:
Taking the columns in turn:
e, f, g -> The RMS should move the Sequence
into closed state.
This is described (in working draft 07)
by lines 759 - 761, and referenced from line 383. I believe that information
is complete and clear, but welcome any suggestions to improve it.

h -> No change, the Sequence is still
closed.
This can easily occur, if the rate that
the RMS sends messages is higher than the time it takes for a fault message
to travel back from the RMD. In any case, I think that the information
noted above covers this, but welcome editorial suggestions.

i -> With the spec as it stands there
is no 'terminating' state for sequences, so we cannot hope to have a sensible
answer here. If a terminating state is created then the RMS may choose
to note the final ack state (as carried in the fault), but should remain
in the terminating state.

j -> The question mark here is covered
by another new issue "Fault Messages for Terminated Sequence",
raised by Tom and recorded as Proposed 02 for the 5th Jan 2006 call.

Given the above, I think this issue should
be closed with no action, but accepting it as an issue may give TC members
time to consider wordsmithing the SequenceClosedFault text.

Cheers,

Matt

Matthew Lovett/UK/IBM@IBMGB

15/12/2005 18:28

To	ws-rx@lists.oasis-open.org
cc
Subject	[ws-rx] State table for RMS

Hi all, 

During the F2F I was asked to start preparing some info on the RMS state
table, particularly when an RMS receives a sequence closed fault, in states
E - I (row 16)? 

In states E, F, G -> Move to closed state 
In state H -> No change, sequence is still closed

I assert that column I should not exist (yet) 

I think this is backed by the current text, if people want to add explanatory
text to the SequenceClosedFault then they should propose that, but I think
that the SequenceClosed fault just told the RMS that the sequence is now
closed.... 

Thanks, 

Matt 

PS: This is based on the spreadsheet here: http://lists.oasis-open.org/archives/ws-rx/200512/xls00001.xls