Re: comments on draft-ietf-sming-reqs-02.txt: Group references

[Juergen Schoenwaelder <schoenw@ibr.cs.tu-bs.de>]

>>>>> David Harrington writes:

David> I think this may have been rejected as unnecessary. 

I do not agree with that. The clauses were specifically added to the
SPPI because they allow code generators to do the right thing. How did
you come to the conclusion they were unnecessary?

David> It is already possible to specify references such as this
David> without having formal notation for it. I believe consensus was
David> that the functionality is a requirement, since it is found in
David> both PIBs and MIBs.

David, this issue is _not_ about having references. It is about a
machine readable notation which allows to express which set of rows a
certain pointer is selecting. This information is extremely important
for code generators as it allows to automate the implementation of
mappings from SMI level references to pointers in your C code.

David> In the quest to keep the language simple, I think the addition
David> of a *formal notation* for this functionality was rejected as a
David> requirement, just as formal notations for many other
David> constraints and relationships were rejected as requirements.

I am unhappy with this kind of logic. I believe decisions should be
made on (a) what does it cost in adding a new feature and (b) what are
the benefits of adding a new feature. In this particular case, the
feature already exists in SPPI I certainly see value in it. (I
generally see lots of value in things that allow us to create code
generators that reduce the amount of low-level SNMP programming we
have to do today.)

/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/>