Lack of Support for Named Levels

The OLE DB for OLAP algorithm for generating column names for the flattened dataset relies on the fact that levels can be named. However, some data sources do not support named levels. For such data sources, the following methods are possible:

They can generate dummy names for the tree levels that they possess. This means that the column names in the set L described in the section “Flattening Algorithm” contain dummy names generated by the provider.
They can generate set L so that there is one column per dimension, as opposed to one column per level.
They can choose to not support flattening.

OLE DB for OLAP strongly recommends that the provider support the dummy name method described above. This makes the provider accessible to a larger class of applications.

You can determine exactly which one of these options a provider chooses by the value of the MDPROP_FLATTENING_SUPPORT property in the DBPROPSET_DATASOURCEINFO property set.

If the value of this property is:

Value	Meaning
MDPROPVAL_FS_ FULL_SUPPORT	Provider supports flattening as described in the section “Flattening Algorithm.”
MDPROPVAL_FS_ GENERATED_COLUMN	Provider supports flattening as described in method 1 above.
MDPROPVAL_FS_ GENERATED_DIMENSION	Provider does not support flattening. For such a provider, calling ICommand::Execute with a language dialect of MDGUID_MDX and asking for the IRowset interface results in an error (E_NOINTERFACE)
MDPROPVAL_FS_ NO_SUPPORT	Provider does not support flattening. For such a provider, calling ICommand::Execute with a language dialect of MDGUID_MDX and asking for the IRowset interface results in an error (E_NOINTERFACE)