Wtrlt: Re: Antw: Re: Q: "opjectClass: top" or not?

["Ulrich Windl" <Ulrich.Windl@rz.uni-regensburg.de>]

(forgot the copy to the list)
I wrote:
>>>>> Michael StrÃder<michael@stroeder.com> schrieb am 22.11.2013 um 20:24 in
>> Nachricht <528FAF5A.3050603@stroeder.com>:
>>> Christian Kratzer wrote:
>>>> very interesting point.
>>>>
>>>> I have a customer with an enterprise application that insists on having
an
>>>> explicit objectClass: top on all entries.
>>>>
>>>> Their developers argue that the rfcs mandate an explicit objectClass:
top
>> on
>>>> all entries.  I argue that the wording in the respective rfcs is not
exact
>>>> enough.  Having an objectClass that inherits from top should be enough
from
>> 
>>> my
>>>> point of view.
>>>>
>>>> I am currently travelling and cannot lookup the rfc.
>>>
>>> I vaguely remember having researched this as well many years ago. Don't
>>> remember the details but I came up with the following recommendations:
>>>
>>> For client developers: Don't require object class 'top'.
>>>
>>> For LDAP admins: Add object class 'top' to all entries.
>>>
>>> Ciao, Michael.
>> 
>> I wonder: If you query an object, you get a set of attributes. Whether you
>> added "top" or not, the attributes are the same (except the "objectClass:
>> top"). And ussually the attributes are not associated with a specific
>> objectlass in the search result.
>> 
>> If you try to "structure" the search results, then with "top", you'll find

> the
>> objectClass attribute at the "top" abstract class, while without "top" 
> you'll
>> find it in the structural object class.
>> 
>> I haven't found a document that describes "multiple inheritance" as it 
> happens
>> with "top" (every objectclass inherits from top, but still the attributes 
> occur
>> only once in the type hierarchy).
>> 
>> Regards,
>> Ulrich