[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: comments on draft-ietf-rap-frameworkpib-02
>>>>> I wrote:
>> Was the resolution not to move the Role and RoleCombination TCs
>> into the FRAMEWORK-PIB? Once they are there, you can apply the
>> PIB->MIB translation rules which gives you a FRAMEWORK-PIB-MIB with
>> Role and RoleCombination TCs that can be imported from the
>> POLICY-DEVICE-AUX-MIB.
>>>>> Keith McCloghrie writes:
Keith> This is correct in the abstract. However, while many users
Keith> have MIB tools, few of them have a tool to do PIB->MIB
Keith> translation. At the same time as they will use the
Keith> POLICY-DEVICE-AUX-MIB, they also need to be supplied with
Keith> whichever MIB contains the Role and RoleCombination TCs. I
Keith> suggest it would be pragmatically advantageous to include the
Keith> Role and RoleCombination TCs in a MIB, in the same document as
Keith> the POLICY-DEVICE-AUX-MIB. While that MIB could be the
Keith> FRAMEWORK-PIB-MIB produced by the PIB->MIB translation of the
Keith> FRAMEWORK-PIB, doing so seems like a lot of overkill. I
Keith> believe it would be a whole lot simpler to just copy the text
Keith> of the two TC definitions into the POLICY-DEVICE-AUX-MIB (as I
Keith> proposed above).
The SPPI includes the PIB->MIB translation rules because you consider
it important to be able to use PIB definitions as MIB modules. Now we
have the first time where this feature is needed and you argue that
the PIB->MIB translation does not work because (a) you can't assume
that people have PIB->MIB translators and (b) translating the whole
PIB is too much work.
There are obvious answers: Either make sure that everyone has access
to a PIB-MIB translator (might not be practical) or make sure the PIBs
are small enough so that the publishing the translated PIB-MIB is not
too expensive. So why do you not put the Role and RoleCombination TCs
in its own small PIB? This addresses issue (b) and you can publish the
PIB and the PIB-MIB in a single document.
An alternative could be to create a single ...-ROLE-MIB which contains
the Role and RoleCombination TCs but nothing else (so the dependency
should not be a problem).
If we copy the definitions by value, then I guess we set a precedence
which leads others to do the same (ignoring the PIB->MIB translation
rules) and we might end up in a mess of copied definitions distributed
over various PIB/MIB modules (which do not even follow a common naming
convention) and it is just a matter of time when it becomes very hard
to enforce consistency.
/js
--
Juergen Schoenwaelder Technical University Braunschweig
<schoenw@ibr.cs.tu-bs.de> Dept. Operating Systems & Computer Networks
Phone: +49 531 391 3289 Bueltenweg 74/75, 38106 Braunschweig, Germany
Fax: +49 531 391 5936 <URL:http://www.ibr.cs.tu-bs.de/~schoenw/>