Re: Thinking about operational attribute numSubordinates...

[Pierangelo Masarati <masarati@aero.polimi.it>]

> Michael Ströder wrote:
> 
> > Pierangelo Masarati wrote:
> > >
> > > On the contrary, a simple boolean isleaf flag, or even a count
> > > of the children, could be computed and returned only if required
> > > with little effort. Comments?
> >
> > I would also be happy with hasSubordinates attribute.
> >
> > http://www.alvestrand.no/objectid/2.5.18.9.html
> >
> > Ciao, Michael.
> 
> I think the "little" effort of computing it on the fly would be too much
> overhead.  It should be maintained as an operational attribute in the
> database.  It could be optional, like maintaining modification times is
> (or used to be).  How would slapcat/slapadd treat this?  Slapcat should
> not dump it, right?

I insist on computing it (only if required and only by those backends
that accept to honor it, maybe configurable, why not).  What scares me
is keeping an entry updated every time its children are added/deleted/
renamed.

I also note that, quoting from the above reported URL:

	"OID description:
	The hasSubordinates operational attribute indicates 
	whether any subordinate entries exist below the entry 
	holding this attribute. A value of TRUE indicates that
	subordinates may exist. A value of FALSE indiciates 
	no subordinates exist. If this atribute is absent 
	no information is provided about the existence 
	of subordinate entries. "

A value of TRUE indicates that subordinates __MAY__ __EXIST__ ...
Sounds like one should not trust it :)

Pierangelo.