Re: Schema for posixGroup successor (RFC 2307 bis)

On Thu, Feb 12, 2015 at 12:17:43PM +0100, Michael Ströder wrote:

> Ok, you've overlooked that I wrote 'groupOfEntries' instead of 'groupOfNames'.

oops... I had assumed you were going for a standalone definition, so I
did indeed misread it.

> In draft-howard-rfc2307bis-02 'posixGroup' is declared as AUXILIARY, a change
> which is forbidden according to Kurt's IANA rules.  And the draft contains a
> declaration of 'groupOfMembers' which is pretty much the same as
> 'groupOfEntries'.  We will sort that out...

Right. That is why I want to separate the groups work from any 2307
updates as far as possible.

> > Looking through the common schemas it seems that multiple inheritance
> > is extremely rare, so I wonder whether all servers would even agree on
> > how it works...
> 
> It works for me. :-)
> 
> Ok, it works in OpenLDAP since years.  We could send a poll ot ldapext and
> ldapbis mailing lists which server products officially support multiple
> inheritance.

That would be useful. Are we agreed about how it *should* work?
My understanding is:

The new class has a set of MUST attributes that is the union of the
MUST attributes of the superior classes.
The new class has a set of MAY attributes that is the union of the
MAY attributes of the superior classes.
Any attribute that is MUST in one superior and MAY in another will
become MUST in the new class.

Andrew
-- 
-----------------------------------------------------------------------
|                 From Andrew Findlay, Skills 1st Ltd                 |
| Consultant in large-scale systems, networks, and directory services |
|     http://www.skills-1st.co.uk/                +44 1628 782565     |
-----------------------------------------------------------------------