Re: [dhcwg] Comment on a couple of option drafts that have gone by...

[Kim Kinnear <[e-mail]>]

At 06:25 PM 8/3/2004, Ted Lemon wrote:
>On Aug 3, 2004, at 1:17 PM, PARK SOO HONG wrote:
>>Isn't it enough to satisfy your concern ?
>
>No, my concern is not related to what you're talking about.   My concern is actually unrelated to the work you are doing - I'm just asking you to add some text to try to get client implements to be consistent in what they do with the parameter request list.   It's not your fault that this is (IMHO) necessary - the problem is that the parameter request list option is poorly specified, and I'd like implementors of this option to implement to the more stringent interpretation of the specification, rather than the more lax implementation.
>
>In order to get client implementors to do this, I am asking you to add some text which will (I hope) cause client implementors to assume that the DHCP server will behave in a certain way.   If they believe that a DHCP server will very likely not send a tep option if the client doesn't request it in the parameter request list, then they will request it in the parameter request list.   If they do not believe this, then they may send a parameter request list but omit the tep option code from the list, thinking it is not necessary to include it.
>
>I don't actually care if server implementors follow the requirements I'm asking you to add - I just care that client implementors have reason to think that servers might follow these requirements.

        Ted,

        Thank you for such a clear statement of this issue.  I
        agree that the parameter request option is poorly
        specified, and I remember my astonishment the first time
        I heard the interpretation that a DHCP server should send
        *every* option with which it is configured (and not just
        the options that were requested in the parameter request
        list).  That wasn't my interpretation of the DHCP
        draft/RFC and it seems it wasn't yours either.

        I also agree with your goal -- that client implementors
        *must* ask (via the paramter request list) for options
        that they need, and not assume that the DHCP server will
        just send send them by magic.

        I believe that the way to achieve that goal is to simply
        put that in the RFC -- that a client that needs this
        option MUST request it via the parameter request list. 

        I strongly believe that saying that a DHCP server MUST
        NOT send this option unless a client requests it is very
        much not the way to achieve this end.  This will require
        special case code in every DHCP server which wants to be
        compliant with this draft (and future RFC).  I see this
        as a "cure worse than the disease".

        Let us just say what we mean -- a client that wants this
        option MUST request it by using the parameter request
        list.  Period.

        Cheers -- Kim



>_______________________________________________
>dhcwg mailing list
>[e-mail]
>https://www1.ietf.org/mailman/listinfo/dhcwg


_______________________________________________
dhcwg mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/dhcwg