Manuel Gaupp wrote:
> Michael Ströder wrote:
>> Please review this draft intended to be published as informational RFC.
> 
> Section 2 ends with the following requirement:
> 
>    LDAP clients displaying a list of entries of these object classes
>    should use mandantory attribute 'cn' to display select lists, hyper-
>    links etc.
> 
> I think this requirement is a bit too specific for such generic object
> classes. In my opinion, default behavior should be to use the attribute(s)
> from the RDN, any other behavior should depend on the specific use case. 

I expected this text to provoke some discussion. ;-)

But I'd like to encourage deployers to think of sufficiently descriptive
common names stored in attribute 'cn' within a given context. Note that
'namedObject' entries are not really usable without auxiliary object class(es)
which limit context and thus the meaningful name space for 'cn'.

> And, think of multivalued RDNs: "cn=namedPolicyEntry+serialNumber=0815" and
> "cn=namedPolicyEntry+serialNumber=4711" would both be listed as
> "namedPolicyEntry" according to this requirement.

I know that some people use DNs like in your example above to make DNs itself
more readable.

But personally I think that multi-valued RDNs really suck because you have to
rename an entry in case one of the characteristic attributes change which is
more effort when maintaing data.
And I saw too many software components failing to handle them correctly.

So one could understand the recommendation in the I-D also to discourage use
of multi-valued RDNs. ;-)

> Section 4 contains the following definition:
>    The OID arc used for the object class defintions is: iso(1) org(3)
>    dod(6) internet(1) private(4) enter-prise(1) stroeder.com(5427)
>    objectClasses(6)
> This does not match the OIDs used in the object class definitions
> (1.3.6.1.4.1.5427.6 versus 1.3.6.1.4.1.5427.1.389.6).

Fixed. Thanks!

Ciao, Michael.

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to