Time/Place

This meeting is a hybrid teleconference and IRC chat. Anyone is welcome to join...here's the info:

Attendees

Agenda

  1. Changes to messaging form (Eric James, Unknown User (escowles@ucsd.edu)).

Minutes

Standup update

 

frank asseg

  • talked with SCC guy about APIs they have available (HPSS, similar to TSM)

  • issue with scape cluster impl - when using the SCAPE API

  • looking to test the setting/reading of properties

  • looking at PR on benchtool transactions

Greg Jansen

  • working on clean shutdown in tomcat, inc. clustered, using ModeShapeEngine to deploy, undeploy, shutdown

  • will return to cluster performance testing after.

Mike Daines

  • reviewing auth use cases and making notes, will ask Qs in chat

Ben Pennell

  • finished PR for transactions in benchtool

  • will try single node tests with transactions turned on

A. Soroka

  • communicating with Stefano Cossu on restrictive type system

  • has paper reviews for OR taking up time today

Agenda item: Changes to messaging form

  • Eric James and Unknown User (escowles@ucsd.edu) are proposing that we make our messages more efficient and clear

  • Observer converts JCR event bus messages to JMS, receives batches (Eventiterator)

  • We currently trigger an event for only the first property for a node within a batch.

  • Proposal: just send a single message w/the node id and not the property

  • We can either report every single property event separately or report a summary of changes

  • Proposal to make it configurable: summary or separate events or both

  • Proposal to add property ids to summary message: include which properties changed from the event iterator (requires no additional JCR lookup)


Question from Stefano Cossu: Is it possible to make it easier to add types to a resource when they incur constraints or allow new properties/children?

  • A. Soroka: Yes, but it will mean that we must order the triples that are received in an update to put type assertions first.
  • Stefano Cossu will create a ticket for this.

 

Question from Stefano Cossu: Is authorization going to be complete for release?

  • Mike Daines: reviewing use cases now. Yes there is a reference implementation based on roles, which are inherited. There is also a XACML reference implementation
  • Greg Jansen: Yes, it is a key functional requirement for release. Talked about the PEP extension point, basic implementation and a possible forthcoming XACML implementation that needs 1 or 2 more sprints of work.

New Actions

  • No labels