[Freeipa-devel] [PATCH] 1035 case sensitivity when calculating indirect members

[John Dennis]

On 07/19/2012 01:35 PM, Rob Crittenden wrote:
> Petr Viktorin wrote:
>> On 07/19/2012 03:07 PM, Rob Crittenden wrote:
>>> Petr Viktorin wrote:
>>>> On 07/17/2012 08:52 PM, Rob Crittenden wrote:
>>>>> When determining whether a member is direct or indirect we were not
>>>>> doing a case-insensitive comparison which led to marking a member as
>>>>> both direct and indirect (in a test case no less).
>>>>>
>>>>> This patch fixes the comparison and the test.
>>>>>
>>>>> rob
>>>>>
>>>>
>>>> When comparing DNs, you should use the DN class, not string
>>>> manipulations: DN(x) instead of x.lower().
>>>>
>>>> How urgent is this? John's DN patch solves this in a much more thorough
>>>> way, maybe it'd be better to just wait for that.
>>>>
>>>
>>> The problem is we're currently reporting incorrect membership. I figured
>>> this would be a short-term fix unless you think the DN patch commit is
>>> imminent.
>>>
>>> rob
>>
>> Okay. Still it makes sense to do the right thing, DN(x) instead of
>> x.lower().
>
> Ok. We can hold onto this patch until we get a better feeling on when
> the DN work will be committed. If it looks to be a while we can commit
> this as a short-term solution, otherwise it'll just get dropped.

Funny, when doing the DN testing I uncovered exactly this problem 
because I was getting different results than what was happening with an 
unmodified master. I worked this out with Rob on the phone I believe. 
Rob was being good by getting a fix in immediately, thank you Rob. But 
my suggestion would be to just pick up the fix with the dn patch which 
will hopefully be soon.


-- 
John Dennis <jdennis at redhat.com>

Looking to carve out IT costs?
www.redhat.com/carveoutcosts/