Keep hash and shrink attr on save as in v0.2 by default

Defaults will change in the next versions
scverse · Jul 3, 2024 · 211fc9c · 211fc9c
1 parent a5926bf
commit 211fc9c
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 10 deletions.
diff --git a/pyproject.toml b/pyproject.toml
@@ -51,7 +51,7 @@ test = [
 
 [tool.pytest.ini_options]
 python_files = "test_*.py"
-testspaths = [
+testpaths = [
     "./tests", # unit tests
 ]
 
@@ -99,4 +99,4 @@ ignore = [
     "E731",
     # allow I, O, l as variable names -> I is the identity matrix, i, j, k, l is reasonable indexing notation
     "E741",
-]
+]
diff --git a/src/mudata/_core/io.py b/src/mudata/_core/io.py
@@ -20,6 +20,7 @@
 
 from .file_backing import AnnDataFileManager, MuDataFileManager
 from .mudata import ModDict, MuData
+from .config import OPTIONS
 
 #
 # Saving multimodal data objects
@@ -34,13 +35,21 @@ def _write_h5mu(file: h5py.File, mdata: MuData, write_data=True, **kwargs):
     write_elem(
         file,
         "obs",
-        mdata.strings_to_categoricals(mdata._shrink_attr("obs", inplace=False).copy()),
+        mdata.strings_to_categoricals(
+            mdata._shrink_attr("obs", inplace=False).copy()
+            if OPTIONS["pull_on_update"] is None
+            else mdata.obs.copy()
+        ),
         dataset_kwargs=kwargs,
     )
     write_elem(
         file,
         "var",
-        mdata.strings_to_categoricals(mdata._shrink_attr("var", inplace=False).copy()),
+        mdata.strings_to_categoricals(
+            mdata._shrink_attr("var", inplace=False).copy()
+            if OPTIONS["pull_on_update"] is None
+            else mdata.var.copy()
+        ),
         dataset_kwargs=kwargs,
     )
     write_elem(file, "obsm", dict(mdata.obsm), dataset_kwargs=kwargs)
@@ -128,13 +137,21 @@ def write_zarr(
         write_elem(
             file,
             "obs",
-            mdata.strings_to_categoricals(mdata._shrink_attr("obs", inplace=False).copy()),
+            mdata.strings_to_categoricals(
+                mdata._shrink_attr("obs", inplace=False).copy()
+                if OPTIONS["pull_on_update"] is None
+                else mdata.obs.copy()
+            ),
             dataset_kwargs=kwargs,
         )
         write_elem(
             file,
             "var",
-            mdata.strings_to_categoricals(mdata._shrink_attr("var", inplace=False).copy()),
+            mdata.strings_to_categoricals(
+                mdata._shrink_attr("var", inplace=False).copy()
+                if OPTIONS["pull_on_update"] is None
+                else mdata.var.copy()
+            ),
             dataset_kwargs=kwargs,
         )
         write_elem(file, "obsm", dict(mdata.obsm), dataset_kwargs=kwargs)

diff --git a/src/mudata/_core/mudata.py b/src/mudata/_core/mudata.py
@@ -576,9 +576,7 @@ def _update_attr(
 
         if not any(attr_changed):
             # Nothing to update
-            # return
-            pass
-            # FIXME
+            return
 
         data_global = getattr(self, attr)
 
@@ -848,7 +846,7 @@ def _update_attr_legacy(
         # No _attrhash when upon read
         # No _attrhash in mudata < 0.2.0
         _attrhash = f"_{attr}hash"
-        attr_changed = self._check_changed_attr_names(attr)
+        attr_changed = self._check_changed_attr_names(attr, columns=True)
 
         attr_duplicated = self._check_duplicated_attr_names(attr)
         attr_intersecting = self._check_intersecting_attr_names(attr)
@@ -862,6 +860,10 @@ def _update_attr_legacy(
                     f"Behaviour is not defined with axis=-1, {attr}_names need to be made unique first."
                 )
 
+        if not any(attr_changed):
+            # Nothing to update
+            return
+
         # Check if the are same obs_names/var_names in different modalities
         # If there are, join_common=True request can not be satisfied
         if join_common: