remove the legacy update()

CHANGELOG.md

@@ -8,6 +8,13 @@ and this project adheres to [Semantic Versioning][].
 [keep a changelog]: https://keepachangelog.com/en/1.1.0/
 [semantic versioning]: https://semver.org/spec/v2.0.0.html
 
+## [0.4.0] (unreleased)
+
+### Changed
+
+- `update()` no longer automatically pulls obs/var columns from individual modalities by default. Set `mudata.set_options(pull_on_update=true)`
+  to restore the old behavior. Use `pull_obs/pull_var` and `push_obs/push_var` for more flexibility.
+
 ## [0.3.4]
 
 ### Added
@@ -144,6 +151,7 @@ To copy the annotations explicitly, you will need to use `pull_obs()` and/or `pu
 
 Initial `mudata` release with `MuData`, previously a part of the `muon` framework.
 
+[0.4.0]: https://github.com/scverse/mudata/releases/tag/v0.4.0
 [0.3.4]: https://github.com/scverse/mudata/releases/tag/v0.3.4
 [0.3.3]: https://github.com/scverse/mudata/releases/tag/v0.3.3
 [0.3.2]: https://github.com/scverse/mudata/releases/tag/v0.3.2

src/mudata/_core/config.py

@@ -1,6 +1,6 @@
 import logging as log
 
-OPTIONS = {"display_style": "text", "display_html_expand": 0b010, "pull_on_update": None}
+OPTIONS = {"display_style": "text", "display_html_expand": 0b010, "pull_on_update": False}
 
 _VALID_OPTIONS = {
     "display_style": lambda x: x in ("text", "html"),

src/mudata/_core/io.py

@@ -28,7 +28,6 @@
 from anndata.compat import _read_attr
 from scipy import sparse
 
-from .config import OPTIONS
 from .file_backing import AnnDataFileManager, MuDataFileManager
 from .mudata import ModDict, MuData
 
@@ -46,22 +45,8 @@ def _is_openfile(obj) -> bool:
 def _write_h5mu(file: h5py.File, mdata: MuData, write_data=True, **kwargs):
     from .. import __anndataversion__, __mudataversion__, __version__
 
-    write_elem(
-        file,
-        "obs",
-        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() if OPTIONS["pull_on_update"] is None else mdata.var.copy()
-        ),
-        dataset_kwargs=kwargs,
-    )
+    write_elem(file, "obs", mdata.strings_to_categoricals(mdata.obs.copy()), dataset_kwargs=kwargs)
+    write_elem(file, "var", mdata.strings_to_categoricals(mdata.var.copy()), dataset_kwargs=kwargs)
     write_elem(file, "obsm", dict(mdata.obsm), dataset_kwargs=kwargs)
     write_elem(file, "varm", dict(mdata.varm), dataset_kwargs=kwargs)
     write_elem(file, "obsp", dict(mdata.obsp), dataset_kwargs=kwargs)
@@ -157,26 +142,8 @@ def write_zarr(
             # zarr_format is not supported in this version of zarr
             file = zarr.open(store, mode="w")
         mdata = data
-        write_elem(
-            file,
-            "obs",
-            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()
-                if OPTIONS["pull_on_update"] is None
-                else mdata.var.copy()
-            ),
-            dataset_kwargs=kwargs,
-        )
+        write_elem(file, "obs", mdata.strings_to_categoricals(mdata.obs.copy()), dataset_kwargs=kwargs)
+        write_elem(file, "var", mdata.strings_to_categoricals(mdata.var.copy()), dataset_kwargs=kwargs)
         write_elem(file, "obsm", dict(mdata.obsm), dataset_kwargs=kwargs)
         write_elem(file, "varm", dict(mdata.varm), dataset_kwargs=kwargs)
         write_elem(file, "obsp", dict(mdata.obsp), dataset_kwargs=kwargs)