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

Re: [libsmi] OID values

On Thu, Dec 02, 2004 at 10:40:15AM +0100, Wijnen, Bert (Bert) wrote:

> COuld we just add it to the MIB review guidelines then?
>  
> > Anyway, your posting is useful for compiler writers. Unfortunately, in
> > 2-3 years we will have lost it again. (So keep a copy around.)
> > 
> and if we add it to mib guidelines we will not loose it.

I doubt it is useful to explain in the MIB review guidelines in length
what in principle is legal and then conclude that you better don't use
this. Instead, it would be more useful to simply document the subset
of legal constructions that usually work:

a) don't use the same descriptor to refer to different things 
   in a MIB module

b) don't use different descriptors to refer to the same thing
   in a MIB module

c) write OID values in the format { <name> <number> } or the format
   { <name>(<number>) <number> } whenever possible

d) generally only refer to descriptors that were introduced by using
   an assignment operator, that is do not refer to a desriptor
   introduced by means of an OID value notation

This list needs editorial work and probably quite some discussion. 
But for MIB authors and reviewers, I believe such BCP rules are more
important than a discussion of the real truth about ASN.1 and SMIv2.
(Such a clarification of the SMIv2 would certainly be a different
 project.)

/js

-- 
Juergen Schoenwaelder		    International University Bremen
<http://www.eecs.iu-bremen.de/>	    P.O. Box 750 561, 28725 Bremen, Germany