Deprecate validate parameter on Dataset

src/mikeio/dataset/_dataarray.py

@@ -227,40 +227,32 @@ def _parse_zn(
                 raise ValueError("zn can only be provided for layered dfsu data")
         return zn
 
-    def _is_compatible(self, other: DataArray) -> bool:
-        """check if other DataArray has equivalent dimensions, time and geometry."""
+    def _is_compatible(self, other: DataArray) -> None:
+        """Check that other DataArray has equivalent dimensions, time and geometry."""
         problems = []
-        assert isinstance(other, DataArray)
         if self.shape != other.shape:
             problems.append("shape of data must be the same")
-        if self.n_timesteps != other.n_timesteps:
-            problems.append("Number of timesteps must be the same")
-        if self.start_time != other.start_time:
+        if self.time[0] != other.time[0]:
             problems.append("start_time must be the same")
-        if not isinstance(self.geometry, other.geometry.__class__):
+        if not isinstance(other.geometry, self.geometry.__class__):
             problems.append("The type of geometry must be the same")
-        if hasattr(self.geometry, "__eq__"):
-            if not (self.geometry == self.geometry):
+        elif type(self.geometry).__eq__ is not object.__eq__:
+            if self.geometry != other.geometry:
                 problems.append("The geometries must be the same")
         if self._zn is not None:
-            # it can be expensive to check equality of zn
-            # so we test only size, first and last element
             if (
                 other._zn is None
                 or self._zn.shape != other._zn.shape
                 or self._zn.ravel()[0] != other._zn.ravel()[0]
                 or self._zn.ravel()[-1] != other._zn.ravel()[-1]
             ):
                 problems.append("zn must be the same")
-
         if self.dims != other.dims:
-            problems.append("Dimension names (dims) must be the same")
+            problems.append("dims must be the same")
 
-        if len(problems) > 0:
+        if problems:
             raise ValueError(", ".join(problems))
 
-        return len(problems) == 0
-
     def _get_plotter_by_geometry(self) -> Any:
         # TODO: this is explicit, but with consistent naming, we could create this mapping automatically
         PLOTTER_MAP: Any = {
@@ -1712,7 +1704,7 @@ def _quantile(self, q, *, axis: int | str = 0, func=np.quantile, **kwargs: Any):
                 qd.name = newname
                 res.append(qd)
 
-            return Dataset(data=res, validate=False)
+            return Dataset(data=res)
 
     # ============= MATH operations ===========
 

src/mikeio/dataset/_dataset.py

@@ -95,12 +95,19 @@ def __init__(
         data: Mapping[str, DataArray] | Sequence[DataArray],
         validate: bool = True,
     ):
+        if not validate:
+            warnings.warn(
+                "validate=False is deprecated and will be removed in a future version. "
+                "Dataset always validates consistency of its DataArrays.",
+                FutureWarning,
+                stacklevel=2,
+            )
+
         data_vars = self._dataarrays_as_mapping(data)
 
-        if validate:
-            first, *rest = data_vars.values()
-            for da in rest:
-                first._is_compatible(da)
+        first, *rest = data_vars.values()
+        for da in rest:
+            first._is_compatible(da)
 
         self._data_vars = data_vars
 
@@ -133,12 +140,21 @@ def from_numpy(
             Geometry of the DataArrays, by default None
         zn: NDArray[np.floating], optional
             Z-coordinates of the DataArrays, by default None
-        validate: bool, optional
-            Validate the DataArrays, by default True
+        validate: bool
+            Deprecated. Dataset always validates consistency.
+            Will be removed in a future version.
         dt: float, optional
             Dummy time step in seconds, by default 1.0
 
         """
+        if not validate:
+            warnings.warn(
+                "validate=False is deprecated and will be removed in a future version. "
+                "Dataset always validates consistency of its DataArrays.",
+                FutureWarning,
+                stacklevel=2,
+            )
+
         item_infos = Dataset._parse_items(items, len(data))
 
         data_vars = {
@@ -148,7 +164,7 @@ def from_numpy(
             for dd, it in zip(data, item_infos)
         }
 
-        return Dataset(data_vars, validate=validate)
+        return Dataset(data_vars)
 
     @property
     def values(self) -> None:
@@ -350,7 +366,7 @@ def fillna(self, value: float = 0.0) -> Dataset:
         """
         res = {name: da.fillna(value=value) for name, da in self._data_vars.items()}
 
-        return Dataset(data=res, validate=False)
+        return Dataset(data=res)
 
     def dropna(self) -> Dataset:
         """Remove time steps where all items are NaN."""
@@ -397,7 +413,7 @@ def squeeze(self) -> Dataset:
         )
         res = {name: da.squeeze() for name, da in self._data_vars.items()}
 
-        return Dataset(data=res, validate=False)
+        return Dataset(data=res)
 
     def create_data_array(
         self,
@@ -562,14 +578,14 @@ def __getitem__(self, key: Any) -> DataArray | Dataset:
                     for k, da in self._data_vars.items()
                     if fnmatch.fnmatch(k, key)
                 }
-                return Dataset(data=data_vars, validate=False)
+                return Dataset(data=data_vars)
             else:
                 item_names = ",".join(self._data_vars.keys())
                 raise KeyError(f"No item named: {key}. Valid items: {item_names}")
 
         if isinstance(key, Iterable):
             data_vars = {v: self._data_vars[v] for v in key}
-            return Dataset(data=data_vars, validate=False)
+            return Dataset(data=data_vars)
 
         raise TypeError(f"indexing with a {type(key)} is not (yet) supported")
 
@@ -677,7 +693,7 @@ def isel(
             )
             for da in self
         ]
-        return Dataset(data=res, validate=False)
+        return Dataset(data=res)
 
     def sel(
         self,
@@ -758,7 +774,7 @@ def sel(
             da.sel(time=time, x=x, y=y, z=z, coords=coords, area=area, layers=layers)
             for da in self
         ]
-        return Dataset(data=res, validate=False)
+        return Dataset(data=res)
 
     def interp(
         self,
@@ -838,9 +854,9 @@ def interp(
                 das = [da.interp(x=x, y=y, interpolant=interpolant) for da in self]
             else:
                 das = [da.interp(x=x, y=y) for da in self]
-            ds = Dataset(das, validate=False)
+            ds = Dataset(das)
         else:
-            ds = Dataset([da for da in self], validate=False)
+            ds = Dataset([da for da in self])
 
         # interp in time
         if isinstance(time, (pd.DatetimeIndex, DataArray)):
@@ -1039,7 +1055,7 @@ def interp_like(
 
         interpolant = self.geometry.get_2d_interpolant(xy, **kwargs)
         das = [da.interp_like(geom, interpolant=interpolant) for da in self]
-        ds = Dataset(das, validate=False)
+        ds = Dataset(das)
 
         if time is not None:
             ds = ds.interp_time(time)
@@ -1222,13 +1238,13 @@ def aggregate(
                 zn=self._zn,
             )
 
-            return Dataset([da], validate=False)
+            return Dataset([da])
         else:
             res = {
                 name: da.aggregate(axis=axis, func=func, **kwargs)
                 for name, da in self._data_vars.items()
             }
-            return Dataset(data=res, validate=False)
+            return Dataset(data=res)
 
     @staticmethod
     def _agg_item_from_items(items: Sequence[ItemInfo], name: str) -> ItemInfo:
@@ -1324,14 +1340,14 @@ def _quantile(self, q, *, axis=0, func=np.quantile, **kwargs) -> Dataset:  # typ
                     geometry=self.geometry,
                     zn=self._zn,
                 )
-                return Dataset([da], validate=False)
+                return Dataset([da])
             else:
                 res: list[DataArray] = []
                 for quantile in q:
                     qd = self._quantile(q=quantile, axis=axis, func=func, **kwargs)[0]
                     assert isinstance(qd, DataArray)
                     res.append(qd)
-                return Dataset(data=res, validate=False)
+                return Dataset(data=res)
         else:
             if np.isscalar(q):
                 res = [da._quantile(q=q, axis=axis, func=func) for da in self]
@@ -1346,7 +1362,7 @@ def _quantile(self, q, *, axis=0, func=np.quantile, **kwargs) -> Dataset:  # typ
                         qd.name = newname
                         res.append(qd)
 
-            return Dataset(data=res, validate=False)
+            return Dataset(data=res)
 
     def max(self, axis: int | str = 0, **kwargs: Any) -> Dataset:
         """Max value along an axis.

src/mikeio/dfs/_dfs0.py

@@ -190,7 +190,7 @@ def read(
             item_infos = [self.items[it] for it in item_numbers]
         else:
             item_infos = self.items
-        ds = Dataset.from_numpy(data, time=ftime, items=item_infos, validate=False)
+        ds = Dataset.from_numpy(data, time=ftime, items=item_infos)
 
         # select time steps
         if self._timeaxistype == TimeAxisType.CalendarNonEquidistant and isinstance(

src/mikeio/dfs/_dfs1.py

@@ -177,7 +177,6 @@ def read(
             time=time,
             items=items,
             geometry=self.geometry,
-            validate=False,
             dt=self._timestep,
         )
 

src/mikeio/dfs/_dfs2.py

@@ -241,7 +241,6 @@ def read(
             time=time,
             items=items,
             geometry=geometry,
-            validate=False,
         )
 
     def append(self, ds: Dataset, validate: bool = True) -> None:

src/mikeio/dfs/_dfs3.py

@@ -268,7 +268,6 @@ def read(
             time=time,
             items=items,
             geometry=geometry,
-            validate=False,
         )
 
     def append(self, ds: Dataset, validate: bool = True) -> None:

src/mikeio/dfsu/_dfsu.py

@@ -306,7 +306,6 @@ def _read_geometry(filename: str) -> GeometryFM2D:
             dfsu_type=dfsu_type,
             element_ids=el_table.ids,
             node_ids=node_table.ids,
-            validate=False,
         )
         dfs.Close()
         return geometry
@@ -486,7 +485,6 @@ def read(
             time=time,
             items=item_infos,
             geometry=geometry,
-            validate=False,
             dt=self.timestep,
         )
 

src/mikeio/dfsu/_layered.py

@@ -180,7 +180,6 @@ def _read_geometry(filename: str) -> GeometryFM3D | GeometryFMVerticalProfile:
             node_ids=node_table.ids,
             n_layers=dfs.NumberOfLayers,
             n_sigma=min(dfs.NumberOfSigmaLayers, dfs.NumberOfLayers),
-            validate=False,
         )
         dfs.Close()
         return geometry
@@ -442,7 +441,6 @@ def read(
                 items=items,
                 geometry=geometry,
                 zn=data_list[0],
-                validate=False,
                 dt=self.timestep,
             )
         else:
@@ -451,7 +449,6 @@ def read(
                 time=time,
                 items=items,
                 geometry=geometry,
-                validate=False,
                 dt=self.timestep,
             )
 

src/mikeio/dfsu/_mesh.py

@@ -58,7 +58,6 @@ def _read_header(self, filename: str | Path) -> GeometryFM2D:
             projection=msh.ProjectionString,
             element_ids=el_table.ids,
             node_ids=node_table.ids,
-            validate=False,
         )
 
         return geom

src/mikeio/dfsu/_spectral.py

@@ -171,7 +171,6 @@ def _read_geometry(
                     dfsu_type=dfsu_type,
                     element_ids=el_table.ids,
                     node_ids=node_table.ids,
-                    validate=False,
                     frequencies=frequencies,
                     directions=directions,
                 )
@@ -184,7 +183,6 @@ def _read_geometry(
                     dfsu_type=dfsu_type,
                     element_ids=el_table.ids,
                     node_ids=node_table.ids,
-                    validate=False,
                     frequencies=frequencies,
                     directions=directions,
                 )
@@ -392,7 +390,6 @@ def read(
             time=time,
             items=items,
             geometry=geometry,
-            validate=False,
         )
 
     def _parse_geometry_sel(

src/mikeio/spatial/_FM_geometry_layered.py

@@ -249,7 +249,6 @@ def to_2d_geometry(self) -> GeometryFM2D:
             projection=self.projection_string,
             element_table=elem_tbl,
             element_ids=elem_ids,
-            validate=False,
             dfsu_type=DfsuFileType.Dfsu2D,
             reindex=True,
         )