I suggest that
the group discuss and agree on the
following:
* What '802.21 compliance'
means? Normative anything would only apply in such a
context.
* Triggers like Link Up, Down etc. are being defined on the MAC SAP
following principles mentioned below. I am still unclear what this MIH layer
really delivers. Before arguing over that SAP being normative or not, the group
ought to agree on the precise definition, interpretation and compatibility
assumptions for the MIH layer.
It does not seem appropriate or necessary
to imply a certain specific implementation strategy for upper layer protocols
(consumers/sources of .21 information). Maybe this is where folks are
confused over what MIH is and isn't. Once that is sorted out, details such as
what the MIH SAP is, what's normative across it etc. should become
obvious.
-Prakash
I was referring to SAP definitions only not APIs. I think we
all agree that we don't
need to standardize APIs. I beleive this
thread was started from a question:
do we need to define higher layer SAP (to
and from MIH) function in .21 WG?
-Subir
Hong-Yon Lach
wrote:
Salut all,
I am lost in the discussion.
In the OSI model and in standardisation, a SAP is part of the abstraction of
a service being standardised. It has nothing to do with how things are
implemented.
Various versions of APIs would be specified for different OS and software
frameworks to facilitate software from different developers to work
together. Typically, the specification and development of APIs are sometimes
based on SAP. Anyway, SAP does not decide how a standard should be
implemented.
Thus, SAP should normally be part of standardisation if its notion is
necessary for the service being standardised. On the other hand API should
not be standardised.
I am aware that the notion of SAP may have been polluted and deviated from
its origin in the OSI model. We need to clarify what we are referring to
here.
Cheers,
Hong-Yon
On 19/01/05 21:51, "Subir Das" <subir@RESEARCH.TELCORDIA.COM> wrote:
I also feel the same way. If we don't need this SAP definition as normative,
then possibly
we don't need the Standardization.
-Subir
Johnston, Dj wrote:
On these grounds, I would support a normative defintion.
Even if the SAP is buried in a mobile device I'm making, I might well be
sourcing things like protocol stack independently. A normative defintion
would lead to a higher liklihood of software component interoperability.
My other reason for preferring a normative defintion is slightly
philisophical.. What is the point of an informative SAP defintion? The
main point of a SAP is both to define a specific service and sometimes
to define the specifics of what information goes over the SAP for
interop purposes. If it's informative it looses both those qualities.
DJ
-----Original Message-----
From: owner-stds-802-21@LISTSERV.IEEE.ORG
[mailto:owner-stds-802-21@LISTSERV.IEEE.ORG] On Behalf Of Cheng Hong
Sent: Wednesday, January 19, 2005 11:16 AM
To: STDS-802-21@LISTSERV.IEEE.ORG
Subject: Re: [802.21] Interface of MIH with Upper layers
Hi Vivek and all,
I think the answer to your question on the SAP depends on whether you
see the implemenators of the MIH and uppper layer the same people. If
they are always the same, maybe a normative definition of the SAP is not
necessary.
However, if there are cases where the MIH and uppper layer (customer of
MIH) are implemented by different people, it probably needs to be well
defined to guarantee interoperability.
cheers
Cheng
-----Original Message-----
From: owner-stds-802-21@LISTSERV.IEEE.ORG
[mailto:owner-stds-802-21@LISTSERV.IEEE.ORG] On Behalf Of Gupta, Vivek
G
Sent: Thursday, January 20, 2005 2:08 AM
To: stds-802-21@LISTSERV.IEEE.ORG
Subject: Interface of MIH with Upper layers
Hello 802.21 Folks,
Another issue which seems to be emerging is the definition of
MIH_L3_SAP or MIH_User_SAP, etc. as some have referred to in their
proposals. The question is should this SAP be defined in normative
sense within 802.21?
This SAP allows higher layer (typically L3) entities to interface with
MIH. This layer to layer communication in a local stack typically
depends on specific OS, driver models in that OS and other
requirements. So what would be the goal and benefits of defining this
interface (SAP) in normative sense? Who would be the consumers of this
and how can we drive/enforce this in normative sense?
Folks have mentioned interoperability and other benefits around this.
But it would be good to get a clear understanding around this as well.
I am not sure if OS abstraction is the goal around some of this(?)
Comments/Thoughts?
BR,
-Vivek
|