• We have agreed on weekly meetings, Friday 11:00-13:00 project (Pacific) time
  
  • This is a short list: 3/3, 3/10, 3/17, 3/24, 3/31, and 4/7 are the remaining Fridays before the 12 Apr 2017 date given us in the charge. We are already aware of some possible impingements on that schedule and we will work around them so that meeting opportunities are not lost.
  • Not everyone is available on 3/10, in particular, so we will attempt to have a ad-hoc meeting in the margins of the JTM.

1. Review charge
2. Some initial issues to work on:
   1. Basic confirmation that something like this is needed
   2. Principles for thinking about Task vs. SuperTask
   3. Contexts in which SuperTask would be used
   4. Patterns of transformations (N-to-M relationships) performed by SuperTask and how the inputs and outputs are to be identified
   5. Naming - the issuance of the WG report in April is the last chance for giving the concept a new name.

Notes:

1) Review charge

Discussion of the "and relevant Butler issues" part of the charge led to the realization that there are other parts of the DM architecture that we'll touch on - for instance, provenance (raised by Jonathan Sick). We agreed that we will not constrain ourselves from pulling in relevant issues to our report, and, just as for the Butler, when we stumble across less tightly-coupled-to-SuperTask issues, we'll ensure that they are notated off to the side for others' future digestion.

2e) Gregory Dubois-Felsmann is not a big fan of the current name but feels the bar is fairly high for finding a new name with sufficiently strong semantics that is also short/pronounceable enough to be used for such a core concept. It should feel comfortable to say "Make a Foo to do X".

• We discussed the question of how to plan for and account for our time at some length.
• A starting point, which Gregory Dubois-Felsmann will report to the T/CAM coord meeting tomorrow, is that we will expect one full day of effort per week from everyone in the group. We agreed that we would suggest capturing this as a 2nd-half-of-S17 epic for each of us. There is a supertask component in DM JIRA already that could be used for this, or, perhaps preferably, we could create a wg-supertask component to isolate the WG effort from Andy Salnikov's (and others') implementation effort.
• Specific implementation work (such as prototype SciPi SuperTasks built by Jim Bosch) could be separately ticketed and would take the effort beyond the 20% level. We were not yet able to define a specific solution for how to know how much of this sort of work would actually arise!

• Michelle Gower gave us an overview of the size and current focus of the NCSA workflow management group.
  • 5-6 people are involved in the effort.
  • They have been evaluating / learning about Pegasus.
  • Now taking a step back to try to firm up the operational requirements for the workflow system.
  • The status/progress of that work is consistent with providing input to the SuperTask WG on the time scale needed.
  • Principal concern is how much the operational workflows will depend on SuperTask capabilities.