Re: [manet] Why the WG Reactive Protocol to be Compatible with another?

["Stan Ratliff (sratliff)" <[e-mail]>]

On Feb 20, 2013, at 5:03 PM, Jiazi Yi wrote:

> Hi, 
> 
> Please check inline (I'm just speaking on behalf of myself, not the LOADng author team) :
> 
> 
> On Feb 20, 2013, at 9:52 PM, Stan Ratliff (sratliff) <[e-mail]> wrote:
> 
>> Jiazi, 
>> 
>> On Feb 20, 2013, at 12:22 PM, Jiazi Yi wrote:
>> 
>>> Hi AB, JP:
>>> 
>>> I also have doubts why the editor of DYMO declares that it will be compatible with LOADng, and how it can be achieved. 
>> 
>> It probably won't be achieved - especially if you and the other LOADng co-authors decide not to help us, taking the position of standing back and lobbing emails like this one into the fray. If this is a desirable, then I would say "suggest some alternatives, help us out". If not, then let us as a working group declare it a dead issue, and move on. 
> 
> In the last 2~years, the LOADng co-authors have spent huge amount of effort in reactive protocols. For me personally, it's the core of my research project, and I invested most of my time in it. Hopefully it won't be regarded as "not willing to help". 

FWIW, here's how I look at it: The past history is what it is. The AD made the decision he made. We are where we are. The operative question is whether or not any given working group participant is willing to help the WG move forward., *from where we are*, to the best reactive protocol we can specify. Any and all assistance and ideas are welcome, and contributors will be acknowledged. 

> 
> 
>> 
>>> 
>>> Although those two protocols share the general idea, the packet format and message processing details are very different. As one who participated in all LOADng interop tests and know dymo draft relatively well, I can say that a lot of effort are needed to make DYMO and LOADng compatible, not to mention the interop tests needed to verify the compatibility (at this point, I have no idea if there is DYMO implementation, so I'm not even sure if such test is possible). 
>> 
>> OK. Going on (yet another) "micro-cosmic trip through MANET reactive protocol history", I'll note that AODV compatibility with LOADng was something that the co-chairs mandated from the combined author teams when we thought that merging the two documents was something that could actually be accomplished. The rationale there was to not exclude existing LOADng implementations from being able to say they were compliant with the RFC - something that *no one* on the author teams (including yourself) complained about *at the time*. 
> 
> Since the Taipei IETF, the starting point agreed by the authors teams was always LOADng, as indicated in Ulrich's mail to the chairs. 
> In that context, all the authors agree that companion documents can be developed, so that the dymo functions can be supported to be compatible. 
> Thanks for adding the note. 
> The issue of starting point didn't come out until several months before Atlanta IETF. If I knew that the document would be based on DYMO, it would be a totally different issue. 

We can quibble about the exact timing of when "the starting point" came out - my memory is that the requirement was clearly stated to the parties involved in YVR. However, my point is that it did come out, has been out "for a while", and no dissent was received. So any implication that this popped up out of the blue is incorrect.

Stan

> 
> 
>> 
>> So to restate, the fundamentals of the issue have changed - the effort to merge the documents was a total, miserable failure. At this juncture it is fair to ask if LOADng compatibility is even something that is desirable. The general trend of comments I'm reading on the list indicates that it is not. Are there other opinions? 
> 
> 
> I agree that the fundamentals of the issue have changed, since the starting point and the design principle (slim core + companion document) were changed. 
> As far as I can see, LOADng has no intention to be changed to "get closer to DYMO". Having DYMO be compatible with LOADng would need great effort and add extra complexity to the protocol -  starting from scratch probably would be easier to make a clear specification and be compatible with LOADng. 
> I have spent days and nights in the interop tests of LOADng. Producing ONE clear specification with limited options, and making it able to interoperable among independent implementations are already hard enough. Having TWO specifications, and one of them is a super set with all those additional options which are not specified clearly, would be a nightmare. Awful among of iterations and tests would be required. 
> 
> Therefore, personally, I would say LOADng compatibility is not desirable. 
> 
> best
> 
> Jiazi
> 
> 
>> 
>> Stan 
>> 
>> 
>>> 
>>> btw, 
>>> 
>>>> the objective of the IETF is to produce qualitative protocols for the best of the IETF community and industry, not to make X or Y happy
>>> 
>>> 
>>> I can't agree more. 
>>> 
>>> best
>>> 
>>> Jiazi
>>> 
>>> On Feb 20, 2013, at 4:26 PM, JP Vasseur (jvasseur) <[e-mail]> wrote:
>>> 
>>>> Hi AB,
>>>> 
>>>> This is critical topic and I cannot agree more with you; the objective is in my opinion to come up with the best protocol for the IETF.
>>>> Let's be clear: the objective of the IETF is to produce qualitative protocols for the best of the IETF community and industry, not to make
>>>> X or Y happy. Every single good ideas borrowed from Load should in my opinion be adopted by AODVv2, if there is a consensus. On the
>>>> other hand, trying to design AODVv2 with the constraint of making it compatible with another protocol is a COMPLETE non sense.
>>>> 
>>>> Thanks.
>>>> 
>>>> JP.
>>>> 
>>>> On Feb 20, 2013, at 3:03 AM, Abdussalam Baryun wrote:
>>>> 
>>>>> Hi Folks,
>>>>> 
>>>>> I am not sure why we will make effort of AODVv2 compatible with
>>>>> LOADng, I am trying to find a good reason mentioned in the WG. I seen
>>>>> the intention by some participants, but not sure was it intention to
>>>>> merge DYMO+LOADng without compatibility (with either DYMO or LOADng)
>>>>> or with percentage of compatibility. Do you think it is good way to
>>>>> make the reactive protocol and then try to make it compatible?  Please
>>>>> advise,
>>>>> 
>>>>> I will have to agree with the below proposal, that if the WG is
>>>>> agreeing to make compatible with LOADng, then why not go for 100%
>>>>> compatibility, just change the name title.
>>>>> 
>>>>> AB
>>>>> 
>>>>> Sub:Re: [manet] Reactive protocol decision process, for the record thoughts
>>>>> On 2/16/13, Ulrich Herberg <[e-mail]> wrote:
>>>>>> Justin,
>>>>>> 
>>>>>> I very much agree with you, Justin. One point I would like to add is
>>>>>> that the WG will now spend a lot of time making the new AODVv2
>>>>>> compatible with LOADng (which is claimed in the current DYMO draft);
>>>>>> that is, in my opinion, a wasted effort since the LOADng draft is per
>>>>>> definition 100% compatible with LOADng (had that been the starting
>>>>>> point).
>>>>>> Ignoring a document that has multiple interoperable implementations,
>>>>>> deployments, MIB document, as well a support from at least a dozen
>>>>>> MANET participants (not only LOADng authors) without giving rational
>>>>>> was a disappointment for me.
>>>>>> 
>>>>>> Best regards
>>>>>> Ulrich
>>>>>> 
>>>>> _______________________________________________
>>>>> manet mailing list
>>>>> [e-mail]
>>>>> https://www.ietf.org/mailman/listinfo/manet
>>>> 
>>>> _______________________________________________
>>>> manet mailing list
>>>> [e-mail]
>>>> https://www.ietf.org/mailman/listinfo/manet
>>> 
>>> _______________________________________________
>>> manet mailing list
>>> [e-mail]
>>> https://www.ietf.org/mailman/listinfo/manet
>> 
>