Re: [ldapext] Schema for posixGroup successor (RFC 2307 bis)

[Simo Sorce <[e-mail]>]

On Wed, 2015-02-11 at 12:24 +0100, Michael Ströder wrote:
> Simo wrote:
> > On Wed, 2015-02-11 at 10:58 +0100, Michael Ströder wrote:
> >> Since Kurt expressed to follow very strict guide lines forbidding altering
> >> existing NAMEs and OIDs of schema descriptions especially in RFC 2307 the
> >> question is how to define a new object class for POSIX groups in RFC 2307bis
> >>
> >> To preserve backwards compability I'd like to propose the following:
> >>
> >> ( <OID TBD>
> >>   NAME 'posixGroup2'
> >>   DESC '<TBD>'
> >>   SUP ( groupOfEntries $ posixGroup )
> >>   STRUCTURAL )
> >>
> >> With this definition..
> >>
> >> 1. posixGroup2 would have 'member' and 'memberUID' both as optional attributes,
> >>
> >> 2. the object class still can be found with (objectClass=posixGroup) and
> >>
> >> 3. the object class is STRUCTURAL and therefore one can assign a specific DIT
> >> content rule to it allowing to preclude either 'member' or 'memberUID' with
> >> NOT to meet local system requirements.
> >>
> >> 4. there's no conflict on LDAP servers with the old RFC 2307 schema already added.
> >>
> >> What do others think about this approach?
> > 
> > I'd call it posixGroupBis, but otherwise sounds like a good idea.
> 
> I have no strong opinion on the NAME.
> 
> > Does it have to be structural though ?
> 
> I expected this question. ;-)
> 
> Yes, it has to be STRUCTURAL and I already gave one rational with 3.  Note
> that DIT content rules, which are very helpful for defining local profiles of
> standard object classes, can only be assigned to STRUCTURAL object classes.
> Also DIT structure rules and name forms only work with this kind of object
> class.  And yes, I'm using all these schema elements. ;-)
> 
> > If it were an auxiliary I might be able to add it to a groupOfEntries
> > object to turn it into a posix group at a later time, that would be
> > nice.
> 
> Yes, I know. But please take note of Kurt's recent statements about IANA
> considerations for NAME 'posixGroup' etc.
> => when using NAME posixGroup anywhere the object class must be STRUCTURAL

Understood, all makes sense.

Simo.