Fix Version/s: None
DM-3472 lands, it will be possible to write an image selection task that uses the SQLite 3 database produced by IndexExposureTask (from daf_ingest) to search for exposures overlapping a region (in particular, the spatial extent of a coadd patch). The _rtree_search method in test_index_exposure.py (also from daf_ingest) has an example of how to perform spatial queries quickly.
I was originally scheduled to do something in this space, but Paul mentioned that he had plans to refactor the image selection tasks already, and is much more familiar with the pipeline side of things than I am. Therefore, I'm handing off the implementation of the pipeline task mentioned in
DM-3472 to him.
Not entirely sure of the status of this ticket but I will note that daf_ingest is currently not part of lsst_apps or lsst_distrib.
- Add spatial database dataset (a sqlite database) in obs_base; will just access by filename.
- Modify ProcessCcdTask to store results in the spatial database.
- Modify SingleFrameDriverTask to store results in the spatial database.
- Modify CoaddBaseTask.selectExposures and the SelectDataIdContainer to use the spatial database.
- Modify CoaddDriverTask to use the spatial database.
Superseded by Gen3 butler. This image selection task will be converted as part of the supertask conversion epic:
Paul Price – I imagine that we'll want to tackle this in the May sprint. I've stubbed in a first guess at roughly how many story points might be involved, but please feel free to adjust that if you think it necessary.