Details
-
Type:
Story
-
Status: Invalid
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_algorithms, pipe_tasks
-
Labels:
-
Story Points:2
-
Team:Data Release Production
Description
SourceDetectionTask should be able to put (at least) an upper bound on the number of post-deblending sources it will put into the SourceCatalog it creates, from the size of the FootprintSet it creates and the total number of peaks within it. It should use SourceCatalog.reserve to ensure the resulting catalog is contiguous.
Attachments
Issue Links
Activity
Is this ticket still valid?
Yup, still valid, and not very hard to do at all. I'll add the PairCoding label to try to get it noticed by somebody looking to find a small ticket to work on at Science Pipelines' weekly pair-coding sesssions.
Bumped into this while working on DM-38546: I'd love to have this fixed soon, so that we don't have to check contiguity in the new CalibrateImageTask.
Update to my comment above: the problem I just ran into was post-deblending. It looks like SourceDetectionTask already pre-allocates correctly using sources.reserve on the output of detectFootprints (see this commit from me in 2018!). Whether the deblender should return a contiguous catalog is a different question, but as-written, I think this is actually done.
I'm not sure ... I've seen reserve not do what I thought it should do, and resize is more reliable. Should we also do a final check? In any event, the deblender should definitely return a contiguous catalog.
As part of the work in
DM-9832, I added some "force-contiguous" tests in calibrate.py and multiBand.py, with TODOs referencing this ticket. Once this is implemented, those can probably go away.