On Mon, Nov 19, 2012 at 12:30:36PM +0100, Pavel Březina wrote:
On 11/16/2012 02:35 PM, Simo Sorce wrote:
>On Fri, 2012-11-16 at 12:22 +0100, Pavel Březina wrote:
>>On 11/16/2012 12:06 PM, Jakub Hrozek wrote:
>>>On Thu, Nov 15, 2012 at 12:55:37PM +0100, Jakub Hrozek wrote:
>>>>On Thu, Nov 15, 2012 at 10:57:37AM +0100, Jakub Hrozek wrote:
>>>>>We broke saving nested LDAP groups with no members in 1.9 during the
>>>>>conversion to ghost users. The attached patches fix that.
>>>>>
>>>>>The first three patches would be nice to get into 1.9, the last patch
is
>>>>>OK in master only. I just found the code hard to read sometimes so I
>>>>>split it into a separate functions and used more descriptive
variable
>>>>>names. I also think that the branch that the branch that re-read
>>>>>SDAP_AT_GROUP_MEMBER after reading SYSDB_GHOST is not needed.
>>>>
>>>>Sorry, self-nack. Changing the conditin was not quite correct in patch
>>>>#2. New set is attached.
>>>
>>>I changed the condition but managed to send patches with the wrong
>>>one..correct set is attached.
>>
>>#1 Ack
>>#2 Ack
>>#3 and #4 Nack
>>>- el->values = talloc_realloc(attrs, el->values, struct
ldb_val,
>>>+ el->values = talloc_realloc(group_attrs, el->values, struct
ldb_val,
>>> cnt);
>>
>>I think el should be used as talloc context instead of group_attrs
>>(ghostel in #4).
>
>No, el is not a talloc_context, it is an element in an array so you
>can't use it. The code above is correct.
OK. You are correct.
Ack to all four patches then.
Thanks.
Pushed to master and sssd-1-9