On the sponsor
ballot I made the following comment..
"In the MAC service primitive definitions, the
CREATE_CONNECTION primitives are replicated as the
CHANGE_CONNECTION
primitives with the following declaration: "The semantics and effect of
receipt
of these primitives are the same as for the corresponding
CREATE
primitives, except that a new CID is not
generated."
The
CREATE_CONNECTION.request does not include a CID parameter, since the response
provides the CID,
as provisioned by the MAC at the time of creation. All
that is in the parameter set of the request is a set of
connection
parameters.
Thus on receipt of a CHANGE_CONNECTION.request, with a new set
of connection parameters, the MAC
somehow has to use the new parameter set
to indentify the CID that is being changed. This is clearly impossible
if
the parameters do not sufficiently match. The CHANGE_CONNECTION.request must
include a CID."
With the loosely
define remedy:
"Write a
definition of the MAC_CHANGE_CONNECTION.request primitive that by duplication
of or reference
to the MAC_CREATE_CONNECTION.request primitive that
defines it to be the same as the
MAC_CREATE_CONNECTION.request primitive
with the exception that an additional paramater 'CID' is included,
to
indicate to the MAC which CID is being changed."
This was rejected
for lack of a specific remedy. I wasn't able to be around for the discussion,
so I have no sense as to whether the group agreed with the comment or whether
some fault in my analysis was identified.
So before I rush
off and write specific text, I'd like to know if people think the comment is
valid.
Thanks,
DJ