Jim Bosch added a comment - 07/Jun/17 8:07 AM Some background reading to catch up with what the SuperTask WG has been doing: SuperTask design requirements: https://confluence.lsstcorp.org/display/DM/SuperTask+design+requirements My perspective on what works and what doesn't in the SuperTask design that tries to meet those requirements: https://confluence.lsstcorp.org/display/DM/SuperTask+Status%3A+Science+Pipelines+Perspective My initial proposal for how to redesign Butler to do what SuperTask needs: https://confluence.lsstcorp.org/display/DM/Data+Units+and+Graph-Based+Inputs+to+defineQuanta (superceded in detail by the PR for DM-10600 , but this may still be useful overview). How the full-scale production system expects to work: https://confluence.lsstcorp.org/display/DM/SuperTask+WG+Input+-+Batch+Processing+Service+High+Level

Pim Schellart [X] (Inactive) added a comment - 12/Jul/17 8:33 AM Hi Jim, I still intend to take some more time to look over the code a second time. But other than that, and excluding possibly talking to other people about this I think I'm reasonably up to speed.

Jim Bosch added a comment - 13/Jul/17 8:01 AM Feel free to close this yourself whenever you think you're done. It's not clear to me whether you want to do more here or not.