The first major task in this epic will involve properly checking through the HSC codebase and deriving a list of functionality which will need to be migrated. Our hope is that Nate Lust will be able to start on that when he arrives in Princeton in early May.

Note that the HSC stack uses its own process distribution middleware which provides functionality not available through LSST. At time of writing, discussions are ongoing about whether it is appropriate to port the HSC middleware to LSST as a stop-gap solution, or to wait until LSST middleware development has ramped up to the extent that it can provide all necessary functionality.

Considerations include:

• Porting the HSC functionality to a new (LSST-developed) middleware would presumably be more effort than continuing to run with HSC-like middleware.
• Development of LSST middleware will be carried out at NCSA, which is currently in the process of staffing up.
• Introducing HSC-like middleware to the LSST stack could create additional work now which will cause our schedule to slip and will ultimately just need to be replaced (and, indeed, its very existence may delay its replacement).
• We regarding merging and re-forking HSC as a technical priority.
• There may be some project-level pressure to process eg DECam data, requiring the functionality now available on HSC.

Discussions are currently ongoing; we should revisit this at the May 2015 DMLT meeting.

The definitive status of middleware will ultimately be a call for the Project Engineer to make, but the noises at the DMLT seemed to indicate that simply merging what's required from HSC would not be a problem in the short term.

Increase the estimated SP total of this epic to 65 at the Summer 2015 mid-cycle reset to reflect the increased scope of work here.