NCL File Processing › Work with Data

Work with Data

As with other NCL functions, data manipulated by file processing is usually maintained in tokenized form (for example, as NCL variables). The &FILE verb also enables data to be maintained in an MDO, but only for mapped format file processing. A typical NCL procedure might accept input from a full-screen panel in a series of tokens or variables and then add this data to a UDB. Alternately, it can receive data from another application using APPC into an MDO, and then add this data to a mapped format UDB.

The physical representation of the data on the UDB depends on its format:

• For mapped format files using $NCL map, the record consists of a list of vectors. Each vector has a two byte length field followed by a two byte tag which is followed by the data from the token. The entire list is encapsulated with its own two byte length and tag fields. For more information about the data structure and tag values, see the distributed $NCL map.
• For unmapped format files, the file processing statements still use variables, but NCL does not regard each variable as a field within the record. When writing a record to an unmapped format UDB all variables are written contiguously (that is, no field separators are inserted). The structure of the data within the record is therefore the responsibility of the user.
• For delimited format files, each variable is treated by NCL as a field within a record, each field being separated from its neighbor by a field separator. When data is retrieved it is returned in a series of variables using the field separators to determine the size of each field. This approach relieves the procedure of having to define the format and structure of data in UDBs.

When data is retrieved from an unmapped format UDB, NCL again makes no attempt to identify individual fields within the record. The entire record is read from the UDB and then split into as many variables as are necessary to hold the data that has been read. Alternatively, the user can indicate how many bytes of data from the record are to be placed into each variable on the &FILE GET statement.

When data is retrieved from a mapped format UDB it is placed into an MDO unchanged from the way it existed on the file. It is also possible to specify the name of a map on the file verb which will be attached to MDO when data is placed into it. When writing to a file using mapped format, the contents of the MDO are simply placed on the file as they are.