Re: [dhcwg] MPL config draft (Re: [Roll] I-D Action: draft-ietf-roll-mpl-parameter-configuration-01.txt)

[Andre Kostur <[e-mail]>]

A couple of quick thoughts (keeping in mind RFC 7227 - Guidelines for
Creating New DHCPv6 Options):

- An encoded option is frowned upon.  Split all of those into separate
options, and the "Reserved" parts simply disappear.  A "sub-optioned"
option is also frowned upon (section 9, RFC 7227).
- Why add the additional complexity of the floating point?   Why not
use a 32-bit unsigned integer representing milliseconds which gets you
a little over 40 days?  Not the 18 weeks that you currently can
represent, but is there a practical use for timers over a month long
in MPL?
- If you do go with separate options, consider what it would mean if
certain options don't exist.  ietf-roll-trickle-mcast appears to
define defaults, perhaps reference that draft to say that if an option
is missing, the client is to use the default defined in that draft?
- Speaking of RFC 7227...  I-D.ietf-dhc-option-guidelines got published in May

-- 
Andre Kostur