Re: LDIF and structural objectclasses

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 02:42 PM 4/23/2003, Daniel Tiefnig wrote:
>if been reading RFC2849 for some time now, and I'm wondering about some
>things.
>
>1) Why are userpassword attributes always encoded in base64 in
>OpenLDAP2.x.x? e.g. "{crypt}c4jHzc08CN.ug" is a valid SAFE-STRING, but
>is encoded anyways.

Because rot13 is not an available LDIF form.  :-)
(See Howard's response.)

>2) I'm not sure that I'm understanding that '*', '1*' and '0*1' thingy
>correctly. AFAICT '*' means as many as wanted and '1*' means as many as
>wanted but at least one. But WTF is '0*1'?

Zero through one.

>3) FAQ-O-Matic says "Object class violation (65)" with "additional info:
>invalid structural object class chain (qmailuser/person)" means "Two or
>more structural objectClass values listed are not related."
>Why's that an error actually? (And since when, have never seen this so far.)

In short, because the information model says each object has
one and only one structure.  In long, see X.501(93).

(This has always been the case in LDAPv3.  2.0 failed to check
for this.  In 2.1, we fixed that bug.)

Kurt