Re: [ipwave] draft-ietf-ipwave-ipv6-over-80211ocb-00 .11-OCB (non) use of beacons

[Alexandre Petrescu <[e-mail]>]

Dear Jérôme,

Le 02/02/2017 à 17:00, Jerome Haerri a écrit :
> Dear Alex,
>
> well, If I cross on red lights, we cannot then say that as I did,
> then crossing on red lights is allowed..

Except that here we are making the red lights too :-)

> if OCB implementations are sending beacons, then it is an
> implementation issue..it does not mean that beacons should be
> transmitted by the standards..

Or vice-versa: the stds may need to reflect what's happening in practice.

In my lab there are such beacons.  In other lab too.  It is for this
reason that I have the doubt.

I can not put in an Internet Draft (I-D) text that simply comes from
other standards.  I-D should contain text that is about running code.

And, of course, I need to put in this I-D what the WG thinks is good.
This is why I am asking.

> Would still suggest to keep the original version or cut-past the
> exact statement from 802.11-2016 about allowed messages in OCB mode.
>
> But just asking frankly: why do you want to keep this statement
> about beacons (allowing them to be sent) ? we spend too much time on
> useless details and not on other key issues..

Because I have no guarantee that if I put there "Beacons MUST NOT be
sent" implementers will do so.

I wonder whether there is a reason why Beacons are there in these OCB
implementations that send them.  Or maybe it's just because they are
forgotten there during migration from .11bgn driver to .11-OCB driver.

Alex

>
> Best Regards,
>
> Jérôme
>
> Envoyé de mon iPhone
>
>> Le 2 févr. 2017 à 16:20, Alexandre Petrescu
>> <[e-mail]> a écrit :
>>
>> But some OCB implementations do send Beacons.
>>
>> In that sense, I propose to say the following.
>>
>>>> 4.  Aspects introduced by the OCB mode to 802.11
>>>>
>>>> In the IEEE 802.11 OCB mode, all nodes in the wireless range
>>>> can directly communicate with each other without
>>>> authentication/ association procedures.  Briefly, the IEEE
>>>> 802.11 OCB mode has the following properties:
>>>>
>>>> o  Wildcard BSSID (i.e., all bits are set to 1) used by each
>>>> node
>>
>> old:
>>>> o  No beacons transmitted
>>
>> new:
>>>> o  OCB implementations do not receive beacons and most OCB
>>>> implementations do not transmit Beacons.
>>
>> Alex
>>
>>
>>> Le 01/02/2017 à 20:20, Dick Roy a écrit : To reiterate, in OCB
>>> operation there are NO beacons and there are NO APs.
>>>
>>>
>>>
>>> RR
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>>
>>>
>>>
*From:*Jérôme Härri [mailto:[e-mail]] *Sent:*
>>> Wednesday, February 1, 2017 4:04 AM *To:* 'Alexandre Petrescu';
>>> [e-mail] *Subject:* Re: [ipwave]
>>> draft-ietf-ipwave-ipv6-over-80211ocb-00 .11-OCB does use beacons
>>>
>>>
>>>
>>> Dear all,
>>>
>>>
>>>
>>> I do not agree and I would be happy to have the reference in the
>>>  IEEE 802.11-2016 stating otherwise.
>>>
>>>
>>>
>>> In the 802.11 standard, it clearly state the following line:
>>>
>>>
>>>
>>> When dot11OCBActivated is true in a STA:
>>>
>>> " The STA _may send management frames of subtype Action_ and, if
>>> the STA maintains a TSF Timer, subtype Timing Advertisement "
>>>
>>> " The STA may send _control frames_, _except_ those of _subtype
>>> PS-Poll, CF-End, and CF-End + CFAck_ "
>>>
>>> " The STA may send _data frames of subtype Data, Null, QoS Data,
>>> and QoS Null_. "
>>>
>>>
>>>
>>> A beacon is a management frame, but not of a class Action, so it
>>>  _cannot be transmitted when dot11OCBActivated_ is true in a
>>> STA. But I agree that it is ambiguous…nevertheless, I would then
>>> replace in the draft the mention of what OCB can do or can’t do
>>> (e.g. sending beacons) with the exact 802.11-2016 statements.
>>> This would avoid misunderstanding.
>>>
>>>
>>>
>>> But even if it could be possible, this does not make much sense.
>>> A beacon is transmitted on a channel you need to connect to get
>>> BSS information and also to get the right frequency. In case of
>>> OCB (outside context of a BSS), it is not required. But the real
>>> question is ‘why’ would we need beacon, assuming it could be
>>> transmitted?
>>>
>>>
>>>
>>> Best Regards,
>>>
>>>
>>>
>>> Jérôme
>>>
>>>
>>>
>>> -----Original Message----- From: its
>>> [mailto:[e-mail]] On Behalf Of Alexandre Petrescu
>>> Sent: Wednesday 01 February 2017 11:51 To: [e-mail] Subject:
>>> [ipwave] draft-ietf-ipwave-ipv6-over-80211ocb-00 .11-OCB does
>>> use beacons
>>>
>>>
>>>
>>> draft-ietf-ipwave-ipv6-over-80211ocb-00 .11-OCB does use beacons
>>>
>>>
>>>
>>> Hello IPWAVErs,
>>>
>>>
>>>
>>> It was suggested in private that, contrary to what the draft
>>> says, the .11-OCB does use beacons.
>>>
>>>
>>>
>>> As such, the following line should disappear from the draft:
>>>
>>>
>>>
>>>> 4.  Aspects introduced by the OCB mode to 802.11
>>>
>>>>
>>>
>>>> In the IEEE 802.11 OCB mode, all nodes in the wireless range
>>>> can
>>>
>>>> directly communicate with each other without authentication/
>>>
>>>> association procedures.  Briefly, the IEEE 802.11 OCB mode has
>>>> the
>>>
>>>> following properties:
>>>
>>>>
>>>
>>>> o  Wildcard BSSID (i.e., all bits are set to 1) used by each
>>>> node
>>>
>>>>
>>>
>>>> o  [strike through] No beacons transmitted
>>>
>>>
>>>
>>> Yours,
>>>
>>>
>>>
>>> Alex
>>>
>>>
>>>
>
>