Don't read full cube into mem for histogramming

docs/changelog.rst

@@ -1,5 +1,16 @@
 Changelog
 =========
+1.0.2
+-----
+Bug fixes
+~~~~~~~~~
+- A full dataset is no longer read into memory when using ``HistogramWidget``.
+  Only the current slice is loaded.
+
+Changes
+~~~~~~~
+- Histogram bin limits are now caclualted from the slice being histogrammed, and
+  not the whole dataset. This is as a result of the above bug fix.
 
 1.0.1
 -----

src/napari_matplotlib/histogram.py

@@ -33,14 +33,19 @@ def draw(self) -> None:
         Clear the axes and histogram the currently selected layer/slice.
         """
         layer = self.layers[0]
-        bins = np.linspace(np.min(layer.data), np.max(layer.data), 100)
 
         if layer.data.ndim - layer.rgb == 3:
             # 3D data, can be single channel or RGB
             data = layer.data[self.current_z]
             self.axes.set_title(f"z={self.current_z}")
         else:
             data = layer.data
+        # Read data into memory if it's a dask array
+        data = np.asarray(data)
+
+        # Important to calculate bins after slicing 3D data, to avoid reading
+        # whole cube into memory.
+        bins = np.linspace(np.min(data), np.max(data), 100)
 
         if layer.rgb:
             # Histogram RGB channels independently