[Freeipa-devel] [PATCH 0042] Removed dead code from LDAPRemoveReverseMember

[Stanislav Laznicka]

On 06/14/2016 04:40 PM, Jan Cholasta wrote:
> On 14.6.2016 16:35, Martin Basti wrote:
>> On 14.06.2016 16:37, Jan Cholasta wrote:
>>> On 14.6.2016 16:29, Martin Basti wrote:
>>>> On 08.06.2016 14:17, Stanislav Laznicka wrote:
>>>>> On 06/07/2016 10:42 AM, Martin Basti wrote:
>>>>>> On 07.06.2016 10:43, Jan Cholasta wrote:
>>>>>>> On 7.6.2016 10:22, Martin Basti wrote:
>>>>>>>> On 07.06.2016 09:07, Jan Cholasta wrote:
>>>>>>>>> On 6.6.2016 18:29, Martin Basti wrote:
>>>>>>>>>> On 03.06.2016 14:28, Stanislav Laznicka wrote:
>>>>>>>>>>> On 06/03/2016 02:19 PM, Martin Basti wrote:
>>>>>>>>>>>> On 03.06.2016 14:13, Stanislav Laznicka wrote:
>>>>>>>>>>>>> https://fedorahosted.org/freeipa/ticket/5892
>>>>>>>>>>>>>
>>>>>>>>>>>> NACK
>>>>>>>>>>>>
>>>>>>>>>>>> please remove it from LDAPAddReverseMember too, it contains 
>>>>>>>>>>>> the
>>>>>>>>>>>> same
>>>>>>>>>>>> code
>>>>>>>>>>>>
>>>>>>>>>>>> Martin^2
>>>>>>>>>>>
>>>>>>>>>>> Please see the modified patch.
>>>>>>>>>>>
>>>>>>>>>>> Standa
>>>>>>>>>>>
>>>>>>>>>> ACK
>>>>>>>>>>
>>>>>>>>>> Pushed to master: c56d65b064e1e0410c03cf1206816cad4d8d86cc
>>>>>>>>>
>>>>>>>>> I think the attrs_list was supposed to be passed to the
>>>>>>>>> ldap.get_entry() call rather than removed, which would fix that
>>>>>>>>> every
>>>>>>>>> reverse member command always acts like --all was specified.
>>>>>>>>>
>>>>>>>> I'm really afraid, what can happen if we put attr_list into
>>>>>>>> get_entry()
>>>>>>>> instead of '*', because this code were there for 4 years and I 
>>>>>>>> don't
>>>>>>>> feel happy enough to change it now, what we may break.
>>>>>>>>
>>>>>>>> Should I revert this commit then and postpone the ticket?
>>>>>>>
>>>>>>> It's a bug and should be fixed. The fix is easy so I see no 
>>>>>>> point in
>>>>>>> postponing it. I see no reason to be really afraid, I'm pretty sure
>>>>>>> that removing the objectclass attribute (which is invisible in the
>>>>>>> CLI anyway) from the output of all the 4 commands that use this 
>>>>>>> code
>>>>>>> won't break anything.
>>>>>>>
>>>>>>
>>>>>> Ok
>>>>> It seems that tests expect objectClass to be always returned in
>>>>> derived methods. Is that expected behavior? If so, please see the
>>>>> attached patch. I wonder if the keys of the passed options should 
>>>>> make
>>>>> it to attrs_list as well (similarly to LDAPUpdate and LDAPCreate)?
>>>>
>>>> I still don't think that we should use that attrs list, because
>>>> according git history, attrs_list was really used only with code that
>>>> was removed, and as you can see Standa had add extra objectclass
>>>> attribute there
>>>
>>> So the assumption here is that if it's in git history, it's not a bug?
>>>
>>> The extra object class should *not* be included by default.
>>>
>>> (Also I don't see any reason to have this split into 2 patches.)
>>>
Right. I added the objectclass so that it behaves similar to what the 
other commands do and so that it does not break tests but it can be 
removed and tests could be fixed. The question here is - do we want it 
in 4.4, then? If so can we be sure it won't break anything (although we 
know that it's broken as it is, and it's been like that for the past 4 
years)?

Currently, the LDAP*ReverseMember are used in adding/removing 
permissions/privileges to privileges/roles so I think we don't want it 
to go bad - but could it?