Fill docs placeholders for old netcdf features.

Author: pp-mo

docs/src/user_manual/explanation/netcdf_io.rst

@@ -133,6 +133,87 @@ Iris' optimisation all together, and will take its chunksizes from Dask's behavi
     (70, 37, 49)
 
 
+Split Attributes
+-----------------
+
+Since Iris does not provide a "dataset" object, it is not possible to represent netcdf
+file-level, aka "global", attributes directly.  Instead, the attributes of *cubes*,
+specifically, are "split" to discriminate between global (i.e. file) and local (i.e.
+variable) attributes.
+
+Prior to Iris v3.8, this was done "automatically" according to a specific lists of
+known attribute names which would normally *only* appear as local (variable) attributes
+(e.g. "valid_min") or *only* as global (file) ones (e.g. "history").
+All other names were saved as "global if the same for all cubes, otherwise local".
+
+Since Iris v3.8, but explicit handling has now been added, so that a ``cube.attributes``
+is a :class:`~iris.cube.CubeAttrsDict` object which combines global and local attributes
+within a single dictionary-like object.
+Finer-grained control of attribute saving is thus now possible, by setting the
+:data:`iris.Future.save_split_attrs` control to ``True``.
+Ideally, anyone who is saving cubes to netcdf files should now set this, and
+a warning is raised if it is not set.
+
+.. note::
+    The default setting remains ``iris.FUTURE.save_split_attrs = False``, purely for
+    backwards compatibility, but it is recommended to override this for all new code.
+
+The :class:`~iris.cube.CubeAttrsDict` provides a regular dictionary access which mimics
+the older single dictionary, while also containing separate lists in its ``.global`` and
+``.local`` properties which allow the user to specifically choose the nature of each
+attribute.
+On saving, each cube's ".attributes.local"s become variable-level attributes and its
+".attributes.global"s become file-level attributes -- *except only*, if globals
+have different values in multiple cubes, then global attributes may be 'demoted' to
+local ones.
+
+Summary
+=======
+
+* You should now always set the ``iris.Future.save_split_attrs=True`` unless you have a
+  specific need for backwards compatibility.
+* all Cube attributes are now stored as specifically "global" or "local"
+* you can continue to assign and fetch cube attributes as before, but when required you
+  can also now access ``.global`` and ``.local`` properties specifically.
+* practically, all this is only relevant when **saving to netcdf files**
+
+
+Deferred Saving
+----------------
+
+In some cases it is useful for performance reasons to defer the actual writing of bulk
+data to a NetCDF file's variables, from after the "iris.save" call to a later time.
+
+You can do this by adding a ``compute=False`` keyword to the ``iris.save`` call.  The
+call then returns, instead of None, a :class:``dask.delayed`` object, which can be
+computed later to complete the writing of variable data to the file.
+
+Where this can be useful is when you are saving multiple cubes whose lazy data is
+computed from shared data, for example multiple statistics over the same data.
+
+To do this, you specify ``compute=False`` to multiple :func:``iris.save`` calls, and
+collect the resulting :class:``dask.delayed`` objects.  You can then complete the saves
+in parallel, by supplying a list of the 'delayed' results to the :func:`Dask.compute`
+function.
+
+This mechanism allows Dask to calculate the multiple "derived" results while only
+fetching the "common" data from which they derive **once**, whereas otherwise the source
+data may need to be fetched multiple times.  In this respect, the benefit is the same as
+the :meth:``iris.cube.CubeList.realise_cubes`` method.  However there is an additional
+potential benefit in this case, which is that the actual **writing** of data to file can
+be performed in parallel.  The scope of this, however, depends on the parallel
+capabilities of the Python ``netCDF4`` package, the NetCDF C library, the file
+system and the operating system :  *Typically* you can write data arrays in parallel
+to **separate files**, but not to different variables, or variable sections *within a
+single file*.
+
+
+Character and String data
+-------------------------
+
+TBC
+
+
 Variable-length datatypes
 -------------------------
 
@@ -189,17 +270,6 @@ loader so it can be make a more informed decision on lazy loading:
     False
 
 
-Split Attributes
------------------
-
-TBC
-
-
-Deferred Saving
-----------------
-
-TBC
-
 .. _save_load_dataless:
 
 Dataless Cubes in NetCDF files