Use epoch argument distinguish multiple instances of the same dataset

doc/source/changelog.rst

@@ -5,6 +5,7 @@ Changelog
 
 X.Y.Z (DD-MM-YYYY)
 ------------------
+* Use epoch to dintinguish multiple instances of the same dataset (:pr:`54`)
 * Use np.logical_or.reduce for generating diffs over more than 2 partitioning arrays (:pr:`53`)
 * Improve Missing Column error (:pr:`52`)
 * Fix `open_datatree` instructions in the README (:pr:`51`)

tests/test_row_mapping.py

@@ -81,7 +81,9 @@ def irregular_ms(tmp_path, request):
 @pytest.mark.parametrize("auto_corrs", [True, False])
 def test_row_mapping(irregular_ms, na, auto_corrs):
   table_factory = TableFactory(Table.from_filename, irregular_ms)
-  structure_factory = MSv2StructureFactory(table_factory, auto_corrs=auto_corrs)
+  structure_factory = MSv2StructureFactory(
+    table_factory, [], "abcdef", auto_corrs=auto_corrs
+  )
   structure = structure_factory()
 
   ddid = table_factory().getcol("DATA_DESC_ID")

tests/test_structure.py

@@ -24,13 +24,14 @@ def test_baseline_id(na, auto_corrs):
 
 
 @pytest.mark.parametrize("simmed_ms", [{"name": "proxy.ms"}], indirect=True)
-def test_structure_factory(simmed_ms):
+@pytest.mark.parametrize("epoch", ["abcdef"])
+def test_structure_factory(simmed_ms, epoch):
   partition_columns = ["FIELD_ID", "DATA_DESC_ID", "OBSERVATION_ID"]
   table_factory = TableFactory(Table.from_filename, simmed_ms)
-  structure_factory = MSv2StructureFactory(table_factory, partition_columns)
+  structure_factory = MSv2StructureFactory(table_factory, partition_columns, epoch)
   assert pickle.loads(pickle.dumps(structure_factory)) == structure_factory
 
-  structure_factory2 = MSv2StructureFactory(table_factory, partition_columns)
+  structure_factory2 = MSv2StructureFactory(table_factory, partition_columns, epoch)
   assert structure_factory() is structure_factory2()
 
   keys = tuple(k for kv in structure_factory().keys() for k, _ in kv)
@@ -84,3 +85,16 @@ def test_table_partitioner():
       },
     },
   )
+
+
+def test_epoch(simmed_ms):
+  partition_columns = ["FIELD_ID", "DATA_DESC_ID", "OBSERVATION_ID"]
+  table_factory = TableFactory(Table.from_filename, simmed_ms)
+  structure_factory = MSv2StructureFactory(table_factory, partition_columns, "abc")
+  structure_factory2 = MSv2StructureFactory(table_factory, partition_columns, "abc")
+
+  assert structure_factory() is structure_factory2()
+
+  structure_factory3 = MSv2StructureFactory(table_factory, partition_columns, "def")
+
+  assert structure_factory() is not structure_factory3()

xarray_ms/backend/msv2/entrypoint.py

@@ -92,7 +92,7 @@ def initialise_default_args(
   epoch = epoch or uuid4().hex[:8]
   partition_columns = partition_columns or DEFAULT_PARTITION_COLUMNS
   structure_factory = structure_factory or MSv2StructureFactory(
-    table_factory, partition_columns, auto_corrs=auto_corrs
+    table_factory, partition_columns, epoch, auto_corrs=auto_corrs
   )
   return epoch, table_factory, partition_columns, structure_factory
 

xarray_ms/backend/msv2/structure.py

@@ -229,16 +229,22 @@ class MSv2StructureFactory:
 
   _ms_factory: TableFactory
   _partition_columns: List[str]
+  _epoch: str
   _auto_corrs: bool
   _STRUCTURE_CACHE: ClassVar[Cache] = Cache(
     maxsize=100, ttl=60, on_get=on_get_keep_alive
   )
 
   def __init__(
-    self, ms: TableFactory, partition_columns: List[str], auto_corrs: bool = True
+    self,
+    ms: TableFactory,
+    partition_columns: List[str],
+    epoch: str,
+    auto_corrs: bool = True,
   ):
     self._ms_factory = ms
     self._partition_columns = partition_columns
+    self._epoch = epoch
     self._auto_corrs = auto_corrs
 
   def __eq__(self, other: Any) -> bool:
@@ -248,16 +254,19 @@ def __eq__(self, other: Any) -> bool:
     return (
       self._ms_factory == other._ms_factory
       and self._partition_columns == other._partition_columns
+      and self._epoch == other._epoch
       and self._auto_corrs == other._auto_corrs
     )
 
   def __hash__(self):
-    return hash((self._ms_factory, tuple(self._partition_columns), self._auto_corrs))
+    return hash(
+      (self._ms_factory, tuple(self._partition_columns), self._epoch, self._auto_corrs)
+    )
 
   def __reduce__(self):
     return (
       MSv2StructureFactory,
-      (self._ms_factory, self._partition_columns, self._auto_corrs),
+      (self._ms_factory, self._partition_columns, self._epoch, self._auto_corrs),
     )
 
   def __call__(self, *args, **kw) -> MSv2Structure: