Meeting began  @ 10:05am; ended 11:50am Project Time.


Next meeting:  @ 10am.


  • Looked at the two new requirements (REQ1103 and REQ1104) and agreed that those are acceptable if they are LSP DAX requirements and not directly Butler requirements. Determined that being able to read a single file made from composites is covered by the requirement to be able to read what a butler writes.
  • Simon Krughoff expressed concerns that the ~100 requirements are unwieldy and we might benefit by having a hierarchy of general requirements and then specific requirements. Unknown User (pschella) and Tim Jennessdidn't think that was a problem but were happy for Simon Krughoff to rearrange the requirements if that helps him.


We had some discussion on 1a vs 1b prioritization and the distinction. We agreed that 1a means the first usable release and 1b means a release 6 months later. We reiterated that the Butler implementation plan will not include the writing of team-specific plugins. There was some question as to how SuperTask and Butler timelines would mesh and Jim Bosch responded that the SuperTask implementation would evolve in parallel with the butler upgrade.

Some questions of priorities were resolved:

  • REQ6362 was lowered to priority 2.
  • REQ669 was lowered to priority 3. There was some confusion over partial datarefs but we decided this was covered in REQ19.
  • REQ775: Lowered to 1b. There was some debate as to whether views should be explicit or whether that was an implementation detail. Simon Krughoff edited the requirement in a more generic way. 
  • REQ882: Handling data quality is critical for operations but we can work around it for now. Lowered to 1b since we won't have a proper data quality system in place in 2018.


Jim Bosch and Unknown User (pschella) presented the current design. There was some commentary on the nomenclature being somewhat different to that used in the requirements (repository vs collection and registry) but we left those differences in place. There was a request to stop using Concrete to mean "in memory" when many people think of the file as being the concrete entity. We had some debate on how metadata is split between the registry and datastore and where provenance might fit. The registry has to have the metadata required to handle data discovery.

Everyone was requested to look at the design and determine how their use cases would fit into the framework.

Use Cases

There was consensus that the Use Cases are now essentially complete and could be migrated to a standalone document. Russell Owen requested that AP1 be looked at by Kian-Tat Lim.

  • Kian-Tat Lim to look at Use Case AP1 and provide feedback  
  • Kian-Tat Lim to look at requirements REQ1999 and REQDAX-95  
  • Tim Jenness to create draft LDM of use cases (LDM-592)  

Working Group End

The stopping criterion for the working group was discussed. We need use cases, requirements, design, decision on whether to use the current butler as a base or start from scratch, and an implementation plan approved by John Swinbank and Fritz MuellerFritz Mueller will talk to Unknown User (npease) to see if he can find time to comment on the requirements one more time next week (and possibly DMTN-056).

  • No labels