Re: SMIng consensus issues restated, call for consensus ends Sept embe r 18, 2002

[Frank Strauss <strauss@ibr.cs.tu-bs.de>]

Hi!

David> You asked for other peoples' reviews. My comments inline.

Thank you!

>> What's the goal of SMING? - Sure, we want to come up a with a new
>> language that makes it
>>  easier to express the model behind a device or service, so that it's
>>  easier to generate better code skeletons for new agent development, and
>>  easier to generate better code stubs for new manager development.
>> 
>> Up to this point, I think we agree.

David> As I recall, we reached the conclusion at the meeting, and in a
David> number of earlier meetings, that the goal of SMING is to make
David> MIB development and use easier for humans, not necessarily for
David> compiler writers. There are very few people who need to write
David> code generators; there are lots and lots of people who need to
David> be able to write and to read mibs.

Yes, it has been mentioned multiple times and I agree. Let me add this:
As long as it does not cost any ease of {read,write}ability, the above
mentioned goals are also worthwhile. 

David> You seem to be of the impression that our goal is to make it
David> easier for code generators. I don't agree that's the general
David> goal. I don't think you'll find many that think that is the
David> goal.

Ok, I must admit that probably this goal is for me more interesting
and more important than to many other people. But I would say, the
human readability, and the clean concepts with less exceptions are not
less important to me. It was my fault not to mention this goals so
explicitly in the above quoted statement.

> [...]

David> I'm not sure what you have in mind that would add nesting to
David> the existing OID hierarchy without changing it.

What I have in mind is nesting in the data structures that people
design.  That's what people need to better express their information
models.  I believe that this not necessarily means to reflect this
nesting when objects are named by OIDs in the protocol. When it comes
to transport instances of them over SNMP, I think it's possible to map
nested data structures, which are named by their identifiers, to OIDs
structured as before in tables. (That's what the NMRG-based proposals
do.)

 -frank