Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: pipe_tasks
-
Labels:None
-
Story Points:4
-
Epic Link:
-
Sprint:DRP F18-4, DRP F18-5, DRP F18-6, DRP S19-1, DRP S19-2
-
Team:Data Release Production
Description
The pipeline task conversion processes involves separating all code which needs access to io from algorithm code. In some cases this is not yet possible (such as with IdFactories) as the middleware necessary for doing this does not yet exist.
However, to make process on the bulk of the conversion process it is possible to use stand in code to allow conversion and testing in place of a final solution.
This ticket will cover the work of going back to all the places where this occurs and fixing it up to a final solution once the required middleware exists.
Attachments
Issue Links
- relates to
-
- Done
Activity
Nate Lust, after all of the excitement over the weekend I forgot that I had basically finished this ticket on Friday (actually, I was waiting for MergeMeasurements to land first, and today realized it didn't actually need any fixing).
In any case, it is now ready for review, and I'll let you decide how you want to stage merging and testing it.
Changes are all in pipe_tasks, and just involve calling packDataId in a few different places: https://github.com/lsst/pipe_tasks/pull/267.
"Only" took three more runs, but I've finally finished testing this with the full suite of PipelineTasks; it all now runs to completion. I've also moved Nate Lust's changes to meas_base from tickets/DM-17496 to this ticket:
https://github.com/lsst/meas_base/pull/136
I've reviewed those, and my only comment is that the commits should be squashed.
I'm starting this now just to fix the IdFactory issue, for which support landed in daf_butler on
DM-14821. There may be some miscellaneous cleanups if I spot anything we've since decided to do differently.