Re: Some additional obscure questions...

["Randy Presuhn" <randy_presuhn@mindspring.com>]

Hi -

> From: "David T. Perkins" <dperkins@dsperkins.com>
> To: "Michael Kirkham" <mikek@muonics.com>
> Cc: <mibs@ops.ietf.org>
> Sent: Thursday, January 30, 2003 7:49 PM
> Subject: Re: Some additional obscure questions...
...
> In ASN.1 modules....

I have no quarrel with what's permitted in ASN.1 modules.

...
> In MIB modules...
>   -- module 1
>   MyModule DEFINITIONS ::=
>   BEGIN
>     IMPORTS
>        ...
>      ;
>      myModule MODULE-IDENTITY
>         ...
>         ::= {0 0 99999 1}
>
>     ...
>     a  ... ::= { foo 1 }
>
>     g OBJECT-GROUP
>        ...
>       ::= { ... }
>   END
>
>   -- module 2 (different from module 1)
>   MyModule DEFINITIONS ::=
>   BEGIN
>     IMPORTS
>        ...
>      ;
>      myModule MODULE-IDENTITY
>         ...
>         ::= {0 0 99999 2}
>
>      ...
>     a  ... ::= { bar 1 }
>     g OBJECT-GROUP
>        ...
>       ::= { ... }
>   END
>
>   AnotherModule DEFINITIONS ::=
>   BEGIN
>      IMPORTS
>         a FROM M1 { 0 0 99999 1 }
>         a FROM M2 { 0 0 99999 2 }
...

As I read it, RFC 2578 doesn't support this.  (RFC 2580 is quite
a different matter.)  Specifically,

|3.2.  IMPORTing Symbols
|
|   To reference an external object, the IMPORTS statement must be used
|   to identify both the descriptor and the module in which the
|   descriptor is defined, where the module is identified by its ASN.1
|   module name.

and in clause 5:

|   Note that reference in an IMPORTS clause or in clauses of SMIv2
|   macros to an information module is NOT through the use of the
|   'descriptor' of a MODULE-IDENTITY macro; rather, an information
|   module is referenced through specifying its module name.

The only mechanism for referencing modules *in IMPORTS* in SMIv2 appears
to be through the module name, not the OBJECT IDENTIFIER, whether
spelled out in { 1 2 3 } form or through the use of the "descriptor".  If the WG
really thinks it's necessary to support this (perhaps for consistency with
RFC 2580), then I suggest that someone should submit an I-D with the
proposed updates to 2578 to make it so.

I was one of the few back in the Dallas interim meeting who wanted to
permit modules to be assigned object identifiers in true ASN.1 fashion,
and still think this would have simplified some versioning issues.  I don't
object to the idea of referencing modules by OID in IMPORTS clauses,
but have a really hard time seeing how the current specification can be
read to support it, especially in view of the history.

Randy