s4:objectclass LDB module - implement additional delete constraint checks

[Matthias Dieter Wallnöfer]

Hi ekacnet,

no I've understood what you mean - and yes, I can bypass the check when 
the "rIDSet" doesn't exist. I will push a patch.

Matthias

Matthieu Patou wrote:
>  Matthias,
>
>
>> Hi ekacnet,
>>
>> Matthieu Patou wrote:
>>> Could it be possible to be a bit less "violent" here ?
>>> This change breaks upgradeprovision in full mode when upgrading at 
>>> least alpha10 but I'm pretty sure that alpha8,9 and 11 are broken 
>>> too ...
>>>
>>> The thing is that old provision do not have the rid_set ... it seems :
>> everything what I implement is stated in MS-ADTS and is tested by me 
>> against Windows Server as far as possible.
>> If you need weaker checks (e.g. in the "objectclass" LDB module) then 
>> please use the RELAX control - this should bypass them.
>>
> No pb to test it again a windows server but you have not to forget 
> that some people (and they are more and more numerous) do not have 
> always a provision with the state of art objects in it (otherwise I 
> won't spend my time on upgradeprovision).
>
> In this particular case you could have made the test a bit different 
> as if the attribute do not exists we are not removing it so it won't 
> hit the test, using the relax control is not the best solution as it 
> is a kind of sledgehammer. Also you can ask yourself and the list of 
> the effect of what you are adding to the samdb code with existing 
> provision.
>
> Matthieu.
>