Joe Touch wrote:
>
> Ron Bonica wrote:
>
>>>Minor addition:
>>>
>>>AND all ICMP programmers MUST NOT parse more than 128 bytes of an ICMP
>>>message unless indicated as valid by a nonzero length attribute.
>>
>>Joe,
>>
>>I am not sure that I agree with that part. I think that the following
>>would be a better statement:
>>
>>"Fully compliant ICMP applications must not parse more that 128 bytes of
>
>
> MUST NOT
>
agree
>
>>an ICMP message unless indicated as valid by a non-zero length
>>attribute. However, in the interest of backwards compatibility, vendors
>>may elect to deploy partially compliant TRACEROUTE applications.
>
>
> MAY...
also agree.
>
> Agreed - where "partially compliant" is as defined in the doc.
>
>
>>Partially compliant TRACEROUTE applications parse the ICMP extension
>>header when a length attribute is lacking, assuming that the original
>>datagram field contains exactly 128 octets. Furthermore, partially
>>compliant TRACEROUTE applications only parse the ICMP extension header
>
> ^^^^^^^^^^^^^^^
>
> MUST NOT parse the ICMP... when ... appear to be incorrect ;-)
>
>
>>when its version number and checksum appear to be correct.
>
>
> Yeah - for those on the list, we'll converge on this nit off-list...
agree again. I will crank out another version of the draft as soon as I
return from travel and run it by you.
-r
>
> Joe
>
>
>>>FWIW, the term 'future' above may be misleading; both requirements apply
>>>to all implementations - existing and future, since an exhaustive search
>>>of existing implementations is infeasible.
>>
>>agree.
>>
>>
>>>Joe
>>>
>
>
_______________________________________________
Int-area mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/int-area
