Merge pull request #1044 from lsst/tickets/DM-45517

DM-45517: No longer check parent when checking compatibility

.pre-commit-config.yaml

@@ -23,10 +23,10 @@ repos:
         name: isort (python)
   - repo: https://github.com/astral-sh/ruff-pre-commit
     # Ruff version.
-    rev: v0.5.4
+    rev: v0.5.5
     hooks:
       - id: ruff
   - repo: https://github.com/numpy/numpydoc
-    rev: "v1.7.0"
+    rev: "v1.8.0rc2"
     hooks:
       - id: numpydoc-validation

python/lsst/daf/butler/_dataset_type.py

@@ -284,10 +284,7 @@ def _equal_ignoring_storage_class(self, other: Any) -> bool:
             return False
         if self._isCalibration != other._isCalibration:
             return False
-        if self._parentStorageClass is not None and other._parentStorageClass is not None:
-            return self._parentStorageClass == other._parentStorageClass
-        else:
-            return self._parentStorageClassName == other._parentStorageClassName
+        return True
 
     def __eq__(self, other: Any) -> bool:
         mostly_equal = self._equal_ignoring_storage_class(other)
@@ -296,6 +293,13 @@ def __eq__(self, other: Any) -> bool:
 
         # Be careful not to force a storage class to import the corresponding
         # python code.
+        if self._parentStorageClass is not None and other._parentStorageClass is not None:
+            if self._parentStorageClass != other._parentStorageClass:
+                return False
+        else:
+            if self._parentStorageClassName != other._parentStorageClassName:
+                return False
+
         if self._storageClass is not None and other._storageClass is not None:
             if self._storageClass != other._storageClass:
                 return False
@@ -309,7 +313,8 @@ def is_compatible_with(self, other: DatasetType) -> bool:
 
         Compatibility requires a matching name and dimensions and a storage
         class for this dataset type that can convert the python type associated
-        with the other storage class to this python type.
+        with the other storage class to this python type. Parent storage class
+        compatibility is not checked at all for components.
 
         Parameters
         ----------

tests/test_datasets.py

@@ -280,7 +280,7 @@ def testCompatibility(self) -> None:
 
     def testOverrideStorageClass(self) -> None:
         storageA = StorageClass("test_a", pytype=list, converters={"dict": "builtins.list"})
-        storageB = StorageClass("test_b", pytype=dict)
+        storageB = StorageClass("test_b", pytype=dict, converters={"list": "dict"})
         dimensions = self.universe.conform(["instrument"])
 
         dA = DatasetType("a", dimensions, storageA)
@@ -291,15 +291,27 @@ def testOverrideStorageClass(self) -> None:
         round_trip = dB.overrideStorageClass(storageA)
         self.assertEqual(round_trip, dA)
 
-        # Check that parents move over.
-        parent = StorageClass("composite", components={"a": storageA, "c": storageA})
+        # Check that parents move over. Assign a pytype to avoid using
+        # object in later tests.
+        parent = StorageClass("composite", pytype=tuple, components={"a": storageA, "c": storageA})
         dP = DatasetType("comp", dimensions, parent)
         dP_A = dP.makeComponentDatasetType("a")
-        print(dP_A)
         dp_B = dP_A.overrideStorageClass(storageB)
         self.assertEqual(dp_B.storageClass, storageB)
         self.assertEqual(dp_B.parentStorageClass, parent)
 
+        # Check that components are checked for compatibility but parents
+        # can be different.
+        parent2 = StorageClass(
+            "composite2",
+            pytype=frozenset,
+            components={"a": storageB, "c": storageB},
+        )
+        dP2 = DatasetType("comp", dimensions, parent2)
+        # Components are compatible even though parents aren't.
+        self.assertFalse(dP.is_compatible_with(dP2))
+        self.assertTrue(dP2.makeComponentDatasetType("a").is_compatible_with(dP_A))
+
     def testJson(self) -> None:
         storageA = StorageClass("test_a")
         dimensionsA = self.universe.conform(["instrument"])

tests/test_parquet.py

@@ -629,6 +629,20 @@ def testWriteSingleIndexDataFrameReadAsArrowTable(self):
         # We check the set because pandas reorders the columns.
         self.assertEqual(set(columns), set(columns2))
 
+        # Override the component using a dataset type.
+        columnsType = self.datasetType.makeComponentDatasetType("columns").overrideStorageClass(
+            "ArrowColumnList"
+        )
+        self.assertEqual(columns2, self.butler.get(columnsType))
+
+        # Check getting a component while overriding the storage class via
+        # the dataset type. This overrides the parent storage class and then
+        # selects the component.
+        columnsType = self.datasetType.overrideStorageClass("ArrowAstropy").makeComponentDatasetType(
+            "columns"
+        )
+        self.assertEqual(columns2, self.butler.get(columnsType))
+
         # Check reading the schema.
         schema = tab2.schema
         schema2 = self.butler.get(