Proposed date: Tuesday August 8, 2017 12:30 pm to 2:00 pm Pacific.
Connection: (Bluejeans TBD)
Suggested audience: middleware pundits, especially @MichelleGower, @ktl, and @gpdf
I’ll be presenting my design sketch for how to use an expanded registry database to solve some SuperTask-related Butler problems (and hopefully some other long-standing Butler/CameraMapper problems). My highest priority is making sure SuperTask WG members understand where we landed on these problems, as this work was done after we had mostly stopped having regular meetings. Devs who have heard rumors of these plans to change Butler APIs may also be interested in getting more details.
This is of course quite relevant to Butler WG activities, but in that context it’s jumping ahead of the use-case and requirements gathering that we’re doing now (which I don’t want to derail). I’ll leave it to @timj to decide if/when the WG should take this up; for now, while Butler WG group members are obviously welcome to attend, this should be considered at most a feasibility study of a design concept I’m guessing we might want to utilize when the Butler WG gets to the design stage.
If anyone wants to read up beforehand, the sketch so far is on the supertaskdev branch of obs_base, in the repodb subpackage. The best entry point is probably the doc string for the RepoDatabase class:
https://github.com/lsst/obs_base/blob/supertaskdev/python/lsst/obs/base/repodb/repodb.py#L26
There is a chance we may move the time earlier to 2:30pm to fit in the old SuperTask WG meeting timeslot, if the RFD time slot does not work for some critical attendees.