Re: [dhcwg] Fwd: [Roll] Request for Review: Fwd: I-D Action: draft-doi-roll-mpl-parameter-configuration-02.txt

[Yusuke DOI <[e-mail]>]

Dear Bernie,

Thank you very much for the review.

(2013-08-13 06:29), Bernie Volz (volz) wrote:

> 1. Please make sure to look at
> http://tools.ietf.org/html/draft-ietf-dhc-option-guidelines. This
> document is being adopted by the DHC WG and is critical to follow.

Thanks, next revision will be with the guideline.

> 2. While it may cost a few octets, I would strongly encourage you to
> use separate fields for some items (such as C_K and DM_K). This will
> cost you one octet but keeps the encoding simpler (flags, C_K, and
> DM_K octets).

Ok, two more octets will make no problem. Though, K>16 seems to be uncommon.

> 3. The new data type (Unsigned Short Floating Point) will likely
> delay implementation in some servers (or require 'manual' encoding).
> If it is critical that this be used (instead of a more traditional
> data type), so be it.

This option is expected to be used in 6lowpan environment and message size matters. Not all DHCP servers are required to implement the options. Manual encoding (over hex) is also not too difficult.

I'll add some implementation details on the next revision.

> 4. As stated in draft-ietf-dhc-option-guidelines, this document does
> specify that multiple instances are allowed (see
> draft-ietf-dhc-option-guidelines-13, section 16). However, it makes
> no statement as to whether the order is significant and that is
> covered in draft-ietf-dhc-option-guidelines-13, section 17 (there
> should be no IMPLICIT or EXPLICIT ordering with multiple options).

Thanks, I'll clarify this. In short, there should be no effect on option order. I think I should explicitly say only one option is allowed for a MPL domain.

> 5. You might document the possible length of the options as there
> seems to be only two choices - X and X+16 (if C=1). Is C even
> necessary? Could the option length (i.e., if MPL domain address is
> present) be used to determine this? (It avoids having to test the
> length if C=1 to make sure it is correct.)

Right. I'll remove the C flag.

> 6. I can say nothing about the utility of this option and whether the
> data contained therein is appropriate as I am not an MPL expert - we
> would defer to the ROLL WG in these regards.
>
> </WG Chair Hat OFF>
>
> - Bernie


Thank you very much and I'll soon submit updated revision.

Yusuke