Re: ABNF of numericoid

["Jim Sermersheim" <jimse@novell.com>]

>numericoid, like the BER
>encoding of an OID, needs to be capable of representing
>any OID value allowed by X.680.

Don't the BER-encoding rules for packing the first two object
identifier components into a single octet ((X * 40) + Y) constrain them
(albeit not constrained to today's constraints)?

Jim


>>> "Kurt D. Zeilenga" <Kurt@OpenLDAP.org> 7/14/03 5:54:08 AM >>>
At 12:45 AM 7/10/2003, Jim Sermersheim wrote:
>All,
>
>In [Models], we define:
>numericoid = number *( DOT number )
>
>This allows an invalid oid like 23.666.<whatever>
>
>Are there usages of numericoid that need to allow these non-valid
oids?
>If not, I think it'd be better to show the proper restrictions on the
>first two numbers.

I think it inappropriate to restrict the syntax of a OID
based upon current assignments and assignment practices.
These are subject to change.   numericoid, like the BER
encoding of an OID, needs to be capable of representing
any OID value allowed by X.680.

For example, 3 is a valid representation of an unassigned
OID. It may one day be assigned.  However, assigned or not,
3 may be a reasonable value of an attribute of syntax OID.

I can think of a number of applications where storing
unassigned OIDs in a directory may be useful.  For example,
in an application which tracked usage of unassigned OIDs in
Internet protocols.

Kurt