Re: FW: New Version Notification for draft-bonica-6man-unrecognized-opt-02.txt

[Brian E Carpenter <[e-mail]>]

On 28/07/2018 06:47, 神明達哉 wrote:
> At Fri, 27 Jul 2018 14:52:05 +0200,
> Fernando Gont <[e-mail]> wrote:
> 
>>> The idea makes sense to me.  One quick comment:
>>>
>>> - I think the draft should be more specific about the option length
>>>   rather than (implicitly) leave it to the implementation.  In fact, I
>>>   don't see why it's not the obvious choice of '0'.
>>
>> While underspecified in the current doc, the option size that has an
>> impact on packet drops. So a node might be able to convey a DO of, say 8
>> bytes to the destination, but not one of, say, 512-bytes.
> 
> Yeah, I imagined that kind of intent, too.  And I wouldn't be opposed
> to making the length variable per se.  But my point is that IMO such a
> discussion should be explicitly included in the draft instead of
> leaving it to reader/implementer's imagination.

I agree, and there's another point that occurred to me, previously
mentioned in an off-list comment:

>> However, if someone
>> sends a long (256 byte) option whose data is all zero, that option
>> will be very compressible, if any hops on the path apply compression.
>> The fact that a compressed packet gets through might conceal the fact
>> that a longer one would fail for some reason.
>> 
>> If you buy that, then the option data should be randomized, which makes
>> it much less compressible.

   Brian