Re: sming wg agenda proposal

[Andy Bierman <abierman@cisco.com>]

At 02:12 PM 11/14/2002 +0100, Juergen Schoenwaelder wrote:

>>>>>> Andy Bierman writes:
>
>Andy> MIB descriptors tend to be hard-to-read, almost random
>Andy> abbreviations, in an attempt to create globally unique
>Andy> identifiers with <= 32 chars.  I want to allow for simpler XML
>Andy> names that are only unique within the scope of their parent
>Andy> element. XML element names can have hyphens and SMI descriptors
>Andy> can't.
>
>SMIng did allow hyphens. SMIng did have class structures where each
>class had its own naming scope - no need for globally unique names.
>Globally unique names were only required in the SNMP protocol
>mappings, to facilitate conversion to SMIv2. Mappings to other
>formalisms could have used the shorter names.

This does not work for SMIv2 MIBs converted to SMIv3, 
because you can't change the descriptors for these MIBs.
I am in favor of changing the descriptor rules for SMIv3
as you suggest.  The optional XML-NAME clause should only
be used if needed.


>We have thrown all this over the board to achieve "simplicity". Now we
>talk about introducing additional names again to facilitate more human
>friendly mappings to XML, which kind of acknowledges that there are
>important mappings of our data structures for non-SNMP purposes...

I never disagreed that MIBs should be useful outside of SNMP.


>/js
>
>-- 
>Juergen Schoenwaelder    <http://www.informatik.uni-osnabrueck.de/schoenw/>

Andy