[PATCH] s4-schema_load: Don't clean in_transaction flag until transaction is really finished

[Kamen Mazdrashki]

On Thu, Nov 11, 2010 at 23:28, Andrew Bartlett <abartlet at samba.org> wrote:
>
> The diff makes it look quite funny, but I think you are correct in this
> patch.  There should be no harm in putting off this schema behaviour
> until the very, very end of the transaction.
>

I thought you did it this way on purpose - this way dsdb_schema_refresh()
may be called to refresh schema cache before transaction commit.
Actually we hit this problem only in dsdb_replicated_objects_commit()
function as it is the only one to use two-phase commit explicitly AFAIK.

And wohoo, with this patch, my simple schema  replication test works both
with classSchema and attributeSchema added :)

-- 
CU,
Kamen