[SSSD] [PATCHES] Collection, INI, Trace, DHASH

John Dennis jdennis at redhat.com
Tue Dec 8 17:59:07 UTC 2009


On 12/08/2009 12:53 PM, Dmitri Pal wrote:
> John Dennis wrote:
>> On 12/08/2009 12:35 PM, Dmitri Pal wrote:
>>
>>> Simo Sorce wrote:
>>>
>>>> On Mon, 2009-12-07 at 09:51 -0500, Dmitri Pal wrote:
>>>>
>>>>
>>>>> 0002 - Dhash improvement. See patch comments. (independent patch)
>>>>>
>>>>>
>>>> I am leaning on nacking this one.
>>>> It is redundant and basically just duplicates code.
>>>>
>>>> The only difference between hash_lookup+hash_delete and hash_remove is
>>>> one less lookup, if that's the point one can simply modify hash_delete
>>>> to return the value, or not modify anything and use the delete_callback.
>>>>
>>>> Duplicating code is not the best way.
>>>>
>>>> Simo.
>>>>
>>>>
>>>>
>>> Yes I looked at this one and was concerned with the same thing but I
>>> could not find a way to make less duplication of the code.
>>> I did not want to modify the existing function since I did not want to
>>> go hunting all the places where it might be used.
>>> But I also was concerned about the duplication as you are and tried to
>>> fund a way to use a common function internally for the two or one wrap
>>> another but I ended up doing what I did since I did not find a better
>>> way of doing things. May be I am missing something.
>>>
>>
>> I think Simo's suggestion of using the delete callback in these cases is
>> the best idea. If you allow returning values from the delete operation
>> you have the potential for inconsistencies. How do you know if value
>> returned is still valid after the delete? It might have been freed by
>> the delete callback. No API changes are necessary and no performance
>> impacts are encountered if you just use the callback mechanism.
>>
>>
>>
> Callback misses the parameter for this to work so that one can determine
> in the callback the condition when the callback is invoked "the delete
> operation" vs "destroy table".
> I need this difference so I hope that this will be added to the callback.

I don't have a problem with adding this flag as a parameter to the the 
callback. However it is an API change, you'll have to search the tree to 
identify any uses of the mechanism and patch those as well.


-- 
John Dennis <jdennis at redhat.com>

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



More information about the sssd-devel mailing list