Re: [Gen-art] Gen-ART review of draft-ietf-mip4-multiple-tunnel-support-12

[Henrik Levkowetz <[e-mail]>]

Hi Vijay,

Thank you for the review!

Comments inline:

On 2015-05-15 17:07, Vijay K. Gurbani wrote:
> I am the assigned Gen-ART reviewer for this draft. For background on
> Gen-ART, please see the FAQ at
> <http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.
> 
> Please resolve these comments along with any other Last Call comments
> you may receive.
> 
> Document: draft-ietf-mip4-multiple-tunnel-support-12
> Reviewer: Vijay K. Gurbani
> Review Date: May-15-2015
> IETF LC End Date: May-25-2015
> IESG Telechat date: Not known
> 
> This document is ready as an Experimental Standard.  A couple of points
> below need addressing, though.
> 
> Major: 0
> Minor: 2
> Nits:  1
> 
> Minor:
> 1/ S4.1, second paragraph: "This extension is a non-skippable extension
>     and MAY be added by the mobile node to the Registration Request
>     message."
> 
>     Two comments.
> 
>     First, does "non-skippable" mean "MUST be present"?

No.  From RFC 5944, https://tools.ietf.org/html/rfc5944#section-1.8:

   The Type field in the Mobile IP extension structure can support up to
   255 (skippable and non-skippable) uniquely identifiable extensions.
   When an Extension numbered in either of these sets within the range 0
   through 127 is encountered but not recognized, the message containing
   that Extension MUST be silently discarded.  When an Extension
   numbered in the range 128 through 255 is encountered that is not
   recognized, that particular Extension is ignored, but the rest of the
   Extensions and message data MUST still be processed.  The Length
   field of the Extension is used to skip the Data field in searching
   for the next Extension.

So non-skippable extensions are extensions which cannot be ignored
if they are encountered but not understood by the receiving node.

>  If so, why "non-
>     skippable"?  At least I have not seen such a phrase in an IETF
>     document before.  My suggestion would be to simply say that "This
>     extension MUST be present and ..."

This is Mobile-IP-specific terminology, and has a well-known and precise
meaning within this area.  It is necessary to specify if an extension is
skippable or non-skippable -- it determines both the numeric range from
which the extension number must be allocated, and how the extension must
be processed.

>     Which brings us to the second comment.
> 
>     "... and MAY be added by the mobile node ..."  This extension is
>     "non-skippable" (?) but a mobile node MAY add it.  If the extension
>     is "non-skippable", then why the MAY?
> 
>     Furthermore, if the intent is for some other entity to add this
>     extension if the mobile node does not (hence the justification of
>     MAY), then you should spell out who this entity is.  (Sort of
>     how you do it in S4.2, second paragraph, which contains almost
>     identical language to above without the qualifying last phrase
>     that informs the reader who will add the extension if not the
>     mobile node.)

This doesn't really follow, given that the meaning of non-skippable is
different than conjectured :-)

> 2/ S4.2, second paragraph: Consider changing the "non-skippable"
>     to "MUST be present" (c.f., above comment).

Again, 'non-skippable' must be used here, as it has a very exact
meaning (different from 'MUST be present') that directly affects
how a message is processed.

> Nits:
> 1/ S2.2, the following sentence does not read well:
>     A mobile node, when it registers multiple bindings with its home
>     agent, each using different care-of addresses, then each of those
>     bindings are given a unique identifier.
> 
>   Suggested text:
> 
>     When a mobile node registers multiple bindings with its home
>     agent, each using a different care-of address, then each of those
>     bindings are given a unique identifier.

Looks good to me.

Thanks!

	Henrik