[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Call for censensus on path forward



Hi Wes,

My (neutral) observation regarding complex filtering-

Why not start with the same expectations for logical operations as described in the
definition of the expExpression object type in the disman expression mib?

Since simple is better when it comes to complex, I would also suggest two levels of
compliance- the agent either does the complete specification for complex filtering, or
does not do any filtering at all.  My concern is that many shadings in between would make
it too difficult for management operations to accommodate variations in filtering rules.

Since I assume that complex filtering would be of interest to applications already using
the disman expression mib, I surmise easier adaptation of complex filtering by such
applications.

Looking forward to the next draft!

Regards,

Mark



Wes Hardaker wrote:

> >>>>> On Mon, 23 Sep 2002 10:10:09 -0400, Dinakaran Joseph <dinakar@us.ibm.com> said:
>
> [Sorry for the delay.]
>
> Dinakaran> A simple logical operations for filtering would be a very
> Dinakaran> good thing.
>
> Ok, so you'd like to be able to use more complex expressions like:
>
> (column1 > 100) && ((column2 < 1 || column3 =~ /some.*thing/ ||
>                     (column8 == 8 && column9 == 9)))
>
> Right?  Based on your statement, and the fact that Carl Kalbfleisch
> and David Perkins submitted allowing similar expressions, I'm going to
> take it that this feature is more desired than not.  I'm really
> interested in hearing from anyone that opposes this level of
> complexity being thrown into the protocol.  As I said at the bottom of
> my last draft, I didn't throw this in because I didn't think people
> would like it since it would impose even more requirements on the
> agent side of things.  Any agent developers want to say something
> negative about complex filtering, otherwise it'll go into my next
> draft (in an easy to parse way [read: nested BER sequences]).
>
> More questions, assuming people want this:
>
> 1) should AND/ORing be MUSTs, SHOULDs or MAY?  (can it be optional to
>    implement)
>
> 2) should there be a max depth or grouping level allowed to the
>    expressions?  IE, can an agent say I won't support the above
>    because it's greater than my max depth level of 2 (since the above
>    has 3 levels of parens).
>
> --
> Wes Hardaker
> Network Associates Laboratories