RE: Question about draft-ietf-hubmib-rfc3636bis

["David B Harrington" <dbharrington@comcast.net>]

Hi,

Wow, I didn't even consider the impact on tools. My bad.
Comments below

> On Mon, Apr 18, 2005 at 11:30:07AM -0700, Randy Presuhn wrote:
> 
> > The minimal test requires three modules:
> >     module A defines X
> >     module B imports X from A
> >     module C imports X from B
> > 
> > If X can actually be used in C, we have a potential solution.
> 

Remember that no module B has yet been found yet - no module imports
these values from A to the WG's knowledge. I suggest the problem be
phrased as:
>     module A defines X
>     module B defines X
>     module C imports X from B


This may be more explicitly a problem for the tools, of course,
because the tools load the definitions from A and load the definitons
from B, and that is likely to fail.

I suppose to work around to this would require obsoleting the original
definition and replacing it with a new distinct definition, but since
the original is wide use, that Is not a very viable approach.

I think Juergen's suggestion would resolve this, but is surely ugly
ugly ugly.

How about keeping the current definitions in the MAU-MIB and defining
one more value in the MAU-MIB that says "go look at this other object
which imports its values from an IANA MIB"? In the BRIDGE-MIB, we have
a widely used pathcost variable that needed the range expanded, so if
the value of the original object is 65535, then a second object with
expanded range should be used. 

We could, of course, revisit the SMI. (Just kidding ...)

David Harrington
dbharrington@comcast.net