Hi guys,
a quick heads up about the current refactoring :
- I'm back into the subtree branch, I don't want to spoil the trunk atm
- I'm currently changing the way we point to subentries in selected
entries, using the subentry's UUID instead of the DN
The first tests I ran on entries with a collective attribute subentry
demonstrated that this approach works. I just checked the Add and Delete
operation atm, changed the subentry cache (the key is the UUID) and I
only had one bump in the road : as the collectiveAttributeSubentries
operational attribute is the one supposed to store the subentry
reference, and as it's defined as a DN syntax in RFC 3672, I had to add
this opAttr before returning the resulting entry in lookup and search.
It was not that easy, but nothing serious though.
I will continue my experimentations with the remaining operations, still
checking the collectiveAttributes operations. Some more tests will be added.
Once it works with the collectiveAttributes, I will move to the ACI and
then the Triggers interceptors.
I didn't defined yet an index to store the selected entries (a <subentry
UUID, selected entries> index) at the moment, so subentry deletion, move
and rename fire a full scan starting at the subentry base, because this
index is just an optimization. I will add it later if everything goes well.
If anyone has some comments, they are very welcome.
--
Regards,
Cordialement,
Emmanuel Lécharny
www.iktek.com
