RE: Problem with slapadd

["Smits.Dolf" <Dolf.Smits@siemens.nl>]

Hm,

The explanation is clear. The effect is probably going to be overwhelming.
I work with Siemens, implementing Siemens DirX, which is a x500/ldap server,
so we never could add entries with a double structural objectclass.
As I did some porting project, I met several times objectclass extensions
where every objectclass is simply made "STRUCTURAL", or where the mix of
structural and auxiliary is so strange, that placement of objects has become
impossible and redefinition was necessary.
(at this moment I'm looking at a radius server schema, who just defined
everything structural).

This enhancement needs a lot of documentation, as nearly everyone who
defined their own objectclass, might run into trouble here.

Just a warning about the imperfection of the LDAP world.

Dolf Smits 

-----Original Message-----
From: Kurt D. Zeilenga [mailto:Kurt@OpenLDAP.org]
Sent: woensdag 6 maart 2002 20:24
To: Anthony Brock
Cc: Open LDAP Software
Subject: Re: Problem with slapadd


At 10:47 AM 2002-03-06, Anthony Brock wrote:
>I am running OpenLDAP HEAD branch and attempting to run the command:
>
># slapadd -c -l testdata.ldif
>slapadd: dn="dc=georgefox,dc=edu" (line=8): no structural object classes
provided
>slapadd: dn="uid=abrock,dc=georgefox,dc=edu" (line=91): invalid structural
object class chain (gfuPerson/ctCalUser)
>#
>
>I am assuming that the "no structural object classes provided" is due to
this being the first entry in the database (i.e., I am NOT creating a
"dc=edu" object). However, I am baffled by the second entry. The schema
which includes gfuPerson and ctCalUser has been working for over a year on
the OpenLDAP 2.0.X release series of code. Has something changed to affect
this in the development code?

Yes, the development code enforces more of the data model
(which is necessary to implement more and more features
of the data model).  In particular, it now checks the
structural object class of each entry.  That is, a entry
can be a 'gfuPerson' or a 'ctCalUser' but cannot be both.
It can be something else which is derived from both, but
then it structurally this something.

Your confusion is likely in that while an entry is
structurally of a one and only one structural object
class, it is said to "belong" to that class and every
class from which that class derives.  That a common
confusion in many object oriented systems which allow
subclassing.

Kurt