RE: comments on draft-ietf-sming-reqs-02.txt: Discriminated Unions vs. Real Unions

["Durham, David" <david.durham@intel.com>]

> -----Original Message-----
> From: David Harrington [mailto:dbh@enterasys.com]
> 
> Does your response mean that the document will not change?
> I believe it should not.
> 
[Dave] I believe the document should not change as far as Unions are
concerned (assuming the WG agrees)... But I think the text could use some
clarification along the lines stated below.

> dbh
> "Durham, David" wrote:
> > 
> > > -----Original Message-----
> > > From: Juergen Schoenwaelder [mailto:schoenw@ibr.cs.tu-bs.de]
> > > Sent: Friday, June 29, 2001 12:08 PM
> > >
> > <snip>
> > >
> > > 51: At first glance, it is somewhat confusing that discriminated
> > >     unions are required while unions are just nice to have. A
> > >     discriminated union is a more precise form of a union 
> and hence
> > >     support for unions will fall out when doing 
> discriminated unions.
> > >     And since discriminated unions are required, the nice to have
> > >     feature of unions will fall out anyway. There is certainly the
> > >     issue whether unions without a discriminator are desirable,
> > >     although arbitrary unions are classified as nice to have here.
> > >
> > >     We are confused (and hungry?)
> > 
> > [Dave] Do you interpret a discriminated union as being able 
> to change the
> > underlying base type of an attribute? It was suggested that 
> a "real" union
> > (in C for example) can change the type as well as the semantic
> > interpretation of an attribute (perhaps even a group of 
> attributes). It was
> > mentioned that supporting a real union will require that 
> the SMIng adds ANY
> > to the supported base types (which is not currently 
> supported by SMIv2 or
> > SPPI). I believe that is why there are two different 
> proposed requirements.
> > One for discriminated unions (based on a semantic selector 
> for a statically
> > typed attribute or group of attributes) vs. real union 
> which may also
> > include a base-type selector.
>