Tickets/DM-19942: ForcedPhotCcdTask in PipelineTask mode needs to accept references from multiple patches #168
Conversation
| containedIds = {0} # zero as a parent ID means "this is a parent" | ||
| for record in refCat: | ||
| if expPolygon.contains(record.getCoord().getVector()) and record.getParent() in containedIds: | ||
| record.setFootprint(record.getFootprint().transform(refWcs, expWcs, expRegion)) |
There was a problem hiding this comment.
This doesn't check for area=0 footprints the way the gen2 path does. Is this due to differences between expPolygon.contains() and self.references.fetchInBox()?
There was a problem hiding this comment.
This is resurrected from an old commit, but from what I researched, if there is no footprint (or its zero area) then there is no coord associated with it (as none can be measured. This will cause getCoord and getVector to create special NaN associated objects. One of these objects is never contained inside any sphere geom regions, therefore this record is skipped.
| """ | ||
| dataRef.put(sources, self.dataPrefix + "forced_src", flags=lsst.afw.table.SOURCE_IO_NO_FOOTPRINTS) | ||
|
|
||
| def getSchemaCatalogs(self): |
There was a problem hiding this comment.
Am I correct that this is a behind-the-scenes method that pipeBase.Task requires?
There was a problem hiding this comment.
Yes, it is related to CmdLineTask (though some gen3 might defer to it, but probably should be updated to reflect the dynamic nature of config names etc.
| catalog.getTable().setMetadata(self.measurement.algMetadata) | ||
| datasetType = self.dataPrefix + "forced_src" | ||
| return {datasetType: catalog} | ||
|
|
There was a problem hiding this comment.
The following methods are no longer documented in the superclass (which is now PipelineTask, CmdLineTask).
There was a problem hiding this comment.
This is a method that is in CmdLineTask and will go away when get get off of gen2
| return {datasetType: catalog} | ||
|
|
||
| def _getConfigName(self): | ||
| # Documented in superclass |
There was a problem hiding this comment.
No longer documented in the superclass.
There was a problem hiding this comment.
This is part of CmdLineTask
| @@ -56,7 +56,6 @@ | |||
| from .forcedMeasurement import * | |||
| from .forcedPhotCcd import * | |||
| from .forcedPhotCoadd import * | |||
| from .forcedPhotImage import * | |||
There was a problem hiding this comment.
This file is staying, it is just no longer used as the base class for ForcedPhotCcd and ForcedPhotCoadd, right?
There was a problem hiding this comment.
Thanks for catching that, it was supposed to go, but in one of the rebase ontos I think it got put back in the tree
| """ | ||
| refWcs = self.references.getWcs(dataRef) | ||
| exposure = self.getExposure(dataRef) | ||
| if psfCache is not None: |
There was a problem hiding this comment.
Is PSF caching not possible in gen3?
There was a problem hiding this comment.
It should be, but it would be done in a different way to this, with some kind of config attribute or something. However, I feel that is outside the scope of this ticket, which was just to orthogonalize the two tasks and make the task work with overlapping patches.
natelust
force-pushed
the
tickets/DM-19942
branch
from
01ec831
to
4ae313c
Compare
ForcedPhotCcd assumed each CCD only overlappend one patch, fix.
natelust
force-pushed
the
tickets/DM-19942
branch
from
4ae313c
to
ef823c7
Compare
No description provided.