Meeting 79 – 30 October 2014

Agenda

  • Actions from meeting 78 (Steve)
  • ICAT development (Steve)
  • ICAT Manager (Christophe)
  • IDS development (Steve)
  • TopCAT development (Wayne)
  • IJP development (Brian R)
  • Installation Script (Leonardo)
  • Issues from ICAT production deployments (Any)
  • AOB (Any)

Notes

Apologies:  Christophe (who tried to join but was unable), Daniel and Alistair.

Present: Shelly, Rolf, Milan, Brian M, Brian R, Kevin, Wayne, Steve, Matt and Tom.

  • Action 78:0 – Christophe has now made a number of releases of the ICAT Manager.
  • Action 78.1 – Wayne has produced some new Balsamiq mockups
  • Action 78.2 – The glassfish configuration for Oracle RAC servers is now part of the setup scripts.
  • Action 75.4 – Andy has submitted Use Cases for TopCAT
  • Action 60.3 – As this concerns the IJP it is being transferred to Brian R.
  • Steve reported that the Rolf had tried ICAT 4.4.0 and that it seems to work – but he had not tested the new parts (import/export and other restful things) .  The code is also being tested for DLS but other testers would be appreciated. The main issue to be resolved is the presence of a “role” attribute in the InvestigationGroup table. Rolf stated that the InvestigationGroup without a role attribute was useless so we must add the attribute or abandon the InvestigationGroup. The new agreement was to add the role and to change the documentation to say that if you want separate the authorization rules from who is actually on the proposal then you can use both InvestigationUser and InvestigationGroup where the InvestigationUser reflects the proposal and the InvestigationGroup is used in authorization rules. Otherwise InvestigationUser can be used for both purposes.
  • Steve then reported on the IDS. He has added support for archives organised by datafile (prompted by DLS needs and which appears to work well) in addition to organisation by dataset. He has also added a new mechanism for handling prepareData (which just keeps the expansion into datafile objects and triggers any restoration from archive) and has simplified the interface for communicating which datasets or datafiles to archive when main storage is running low.  There is currently a security problem which has been explained on the icatgroup list. The problem is that the location field cannot be trusted. There are two suggested ways to address this – passing more information to the plugin to permit sanity checks to be made by the plugin or adding a cryptographic hash to the location field – or both. It has also been proposed to not pass the dataset.location via the plugin interface. Tom thought that it was probably best to include both solutions and nobody objected to the elimination of dataset.location from the plugin interface.
  • Wayne reported that TopCAT 1.12 had been released and that yesterday a big update to the balsamiq mock-ups had been announced. Not surprisingly people had not yet studied them. Steve emphasised the importance of making sure that these mock-ups appear to reflect what people really want to see.
  • Brian R reported that he had  – refactored lsfbatch to use shared classes which ensure that all the IJP batch systems have a common interface. He is now considering how to select and pass datafiles (as well as datasets) to jobs. Wayne had suggested that it might be worth using balsamiq to produce a mock-up of the revised GUI.
  • Tom said that they were trialling an “auto-reduction” system with  users next week.

Actions

ITEM DESCRIPTION ASSIGNED
74.0 Look into JDBC connector resilience with Wildfly Steve
61.1 Put the simple ICAT browser into contrib when it is ready Condie
60.3 Finalise a release of IJP and send a note to the mailing list Brian R