Post-review fixups.

Improve documentation/comments on usage of "checkAllCcds".
lsst · Aug 29, 2019 · 33c027e · 33c027e
1 parent 245f2b3
commit 33c027e
Showing 1 changed file with 18 additions and 2 deletions.
diff --git a/python/lsst/fgcmcal/fgcmBuildStars.py b/python/lsst/fgcmcal/fgcmBuildStars.py
@@ -117,7 +117,9 @@ class FgcmBuildStarsConfig(pexConfig.Config):
         default=13,
     )
     checkAllCcds = pexConfig.Field(
-        doc="Check all CCDs.  Necessary for testing",
+        doc=("Check repo for existence of all possible ccds, in case when we have "
+             "incomplete visits and the referenceCCD will be missing.  Primarily "
+             "used in testing."),
         dtype=bool,
         default=False,
     )
@@ -483,14 +485,28 @@ def findAndGroupDataRefs(self, butler, dataRefs):
         # will be unnecessary with Gen3 Butler.  This should be part of
         # DM-13730.
 
+        # Note that self.config.checkAllCcds means that we will check the repo
+        # for existence of all possible ccds for each (incomplete) visit, in cases
+        # when the referenceCCD will be missing (primarily during testing).
+
+        # scanAllCcds is a local flag that is set when the dataRefs do not include
+        # all relevant ccds.
+
         scanAllCcds = False
         if len(dataRefs) == 0:
             if not self.config.checkAllCcds:
+                # These dataRefs include all possible visits, for each
+                # referenceCCD.  This is much faster than scanning through
+                # all CCDs when none of them have been processed in a given repo.
                 dataRefs = butler.subset('src',
                                          dataId={self.config.ccdDataRefName:
                                                  self.config.referenceCCD})
                 scanAllCcds = True
             else:
+                # These dataRefs include all possible ccds
+                # Note that in the Gen2 era (at least) a valid dataRef in a subset
+                # only says that a given raw is available, and the src may not
+                # be accessible or processed in a specific repo.
                 dataRefs = butler.subset('src')
 
         groupedDataRefs = {}
@@ -501,7 +517,7 @@ def findAndGroupDataRefs(self, butler, dataRefs):
                 continue
             # If we need to scan all ccds, do it here
             if scanAllCcds:
-                dataId = dataRef.dataId
+                dataId = dataRef.dataId.copy()
                 for ccdId in ccdIds:
                     dataId[self.config.ccdDataRefName] = ccdId
                     if butler.datasetExists('src', dataId=dataId):