This section presents ADI from the view of the user, namely applications programmers and amateur programmers. The various concepts of the system are explained in an abstract form, the nitty gritty of the implementation is discussed in the next section.
Asterix analysis applications communicate with data residing in the file system using (mostly) subroutine interfaces which provide data abstraction, ie. the information regarding the format used in the underlying file is coded in the access layer rather than the application. This is only good computing practice -- applications are insulated from changes in file formats. When a system evolves using only one data format it is not surprising that the elements of the interface layer map very closely onto components of the underlying data format, and so it is easy to confuse the data model with the data representation.
When new data formats are added one has to ask whether there is a fundamental change required in your data model. The answer for the OGIP FITS formats is clearly not. The requirements driving the definition of the OGIP FITS formats were largely the same as those driving the Asterix extension to the Starlink NDF format, and the differences largely reflect the flat vs. hierarchical file formats preferred in the two cases.
The task of constructing a unified abstract data interface is then one of defining an appropriate data model, and the mappings between the various file formats and that model.