Re: text format of configurations

[Jon Saperia <saperia@jdscons.com>]

>     > User interface has generally been a space 
>     > that the IETF stayed out of.
> 
> I agree.  If vendors come up with uniform syntax themselves, there's no
> reason for anyone else to become involved.  I think that would make
> everyone happiest.  I hadn't been thinking of an IETF document specifying
> syntax, but I can imagine some sort of standards document specifying
> verbose namespace, equivalent to OIDs.  And even equivalencies, for
> multiple-human-language-support and so forth.  Um, I'd better shut up now,
> before I get too far into bad-idea-fairyland.  :-)
> 
>     > 2) If there are new features, there is no way that you can expect multiple 
>     > vendors to create the exact same syntax for controlling these 
>     > features.  This implies that even if everyone buys into the goal of 
>     > uniformity, it will not be perfectly achieved.
> 
> My thought on that was that something kind of like the OID vendor space
> (sorry, I know I don't know SNMP terminology) but verbose, and with a
> long-term pressure toward convergence, was what might be needed.  For
> example, any vendor could come up with any new command that they liked, so
> long as it was prefixed by the string "vendor-experimental-".  As soon as
> they could convince their fellow vendors to standardize on it, they could
> pull that prefix off.  Ugly, but only for the vendors who don't converge.
> 
>                                 -Bill
> 
> 

Bill, Joel and I made similar comments about vendors historic
unwilingess to do the right thing here. That said, I like your
suggestion. Let me build on it a bit. I have build several systems with
command line interfaces similar to those everyone is familiar
with. Forgetting how for the moment, we created commands and SNMP
objects that were related (saved some code development). The result was
that one could register the CLI commands, in fact they really already
were by virtue of being related to OIDs. It thus made it possible to
'map' the cli commands. A side benefit was that we ensured consistency
between the CLI and SNMP world - some people here may see that as a
feature, others a bug. So to use your example, if a new command needed
to be developed say because of a new technology, then we would register
it and perhaps a new OID for a MIB object. I expect that generally many
are needed at a time as new technologies come along or evolve.


> 
> 

Thanks,
/jon
--

Jon Saperia		     saperia@jdscons.com
			     Phone: 617-744-1079
			     Fax:   617-249-0874
			     http://www.jdscons.com/