The data format container described here is too unwieldy to be expected to be used directly in most APIs. The expectation is that APIs and users will define data descriptors in memory, but have API-specific names for the formats that the API supports. If these names are enumeration values, a mapping can be provided by having an array of pointers to the data descriptors, indexed by the enumeration. It may commonly be necessary to provide API-specific supplementary information in the same array structure, particularly where the API natively associates concepts with the data which is not uniquely associated with the content.

In this approach, it is likely that an API would predefine a number of common data formats which are natively supported. If there is a desire to support dynamic creation of data formats, this array could be made extensible with a manager returning handles.

Even where an API supports only a fixed set of formats, it is flexible to provide a comparison with user-provided format descriptors in order to establish whether a format is compatible.