I/O Data in DDS applications are controlled by the use of "alias" names (not to be confused with "alias" Definitions) in identifying each input and output dataset. The "alias" name generally specifies the Dictionary associated with a dataset. These Dictionaries specify the format and binary data filename. The user can override each dataset's "format" and "data" (binary filename) Definition. However, if you define "format" or "data" directly, then that would override those Definitions for each input dataset that might be associated with the application. (Likewise, any command line or par file parameter like "axis=" or "delta.t=" would apply to each input dataset to the application.)

So to get around this DDS uses a modified Definition name for each dataset base on each alias name (eg. alias_format). So for example if you have an input dataset identified as "in", you can override it's format by specifying "in_format=". For example a full specification of an input and output dataset might look like:

in= line34
in_data= Hline34_@
in_format= ccube
out= line34_migrated
out_data= Hline34_migrated_@
out_format= fcube
    

If you look at a created dataset's Dictionary you may see this override Definition (eg. "in_format=fcube") that came as an input parameter followed by DDS creating an alias Definition for the format (eg. "$format=in_format"). The next time DDS tries to open up this dataset, it will look backwards through the Dictionary for "format=". When it finds "$format=in_format" it will then change it's search to "in_format=" and continue searching backward from that point.