Participants
------------

* Sergio Andreozzi
* Stephen Burke
* David Horat
* Balazs Konya
* Maarten Litmaath
* Paul Livesey

Minutes (please send corrections)
-------

* Handling of Entity: agreement to keep Entity prefix for all attributes except
  for the ID, which takes the name of the first derived class as prefix.

* Attributes otherwise keep their names also in derived classes.

* IDs are globally unique (also across different classes).

* Sergio: LDAP objects need 1 abstract class, 1 structural class and
  any number of auxiliary classes.

* Relationships: imitate v1.x "ForeignKey: SomeForeignAttribute=value" syntax
  (no) or David's proposal "SomeForeignAttribute: value" (yes).

* Balazs: copy LDAP DIT from XML rendering.  LDAP also needs to allow for
  grouping elements.

* Stephen: is grouping optional?

* Sergio: in XML mandatory where defined.

* Stephen: DIT not needed for queries.

* Sergio: v1.x uses foreign keys because of LDAP limitations.
  For XML keys are only added where the tree relationships are insufficient.

* David: makes queries difficult.

* Balazs: some relationships are less important, therefore not rendered.

* Stephen: every relationship needs a foreign key.  Agreed.

* Sergio: chunk key was invented for parent-child relationships.

* Stephen: now all IDs are unique, no more local IDs, no special treatment
  needed for parent-child relationships.

* David: should a tree be imposed?

* Stephen: BDII changes the tree at the root!  How many Admin Domains?

* Stephen: some objects are optional --> nodes may be missing.
  DataStore <--> Manager; Activity <--> {Endpoint, Resource, Share}

* David: XML has the same problem?

* Balazs: yes.

* David: imposing a tree conflicts with optional nodes in the middle.

* Stephen: one would have to relate all affected classes to Service etc.

* Maarten: impose structure at lower levels, look at higher levels later.

* Sergio: postpone discussion until we have looked at XML example.

* Stephen: which side of a relationship holds the foreign key?

* David: the side with fewer attributes, else depends on the case.
  We need to look at the undecided cases to try and optimize for queries
  corresponding to expected use cases.

* Next meeting planned for Tue. May 26 at 16:00.

Actions
-------

* David will look at XML rendering and convert example to LDAP.