Users who want to create their own table and variable definitions can do so from the following sources of information :

• from an existing SAS data set
• from an existing table in a PDB
• from a recognized file format (for example, a Class Specification File for HP MeasureWare's Data Source Integration).

They can use the GENERATE SOURCE statement of the %CPDDUTL macro to build a first version of the CREATE TABLE, UPDATE TABLE, and CREATE VARIABLE statements needed to define the table. The generated source statements can be stored in either a SAS ® catalog member of type .SOURCE or in an external flat file.

Typically, a user would use GENERATE SOURCE when setting up to use a generic collector, in order to aid in defining the table to the PDB. As such, the data set used for staging the data for the generic collector would serve as the DATASET= parameter for the GENERATE SOURCE statement.

If you are required to use a staged data set as input to %CxPROCES you must ensure that it conforms to the following rules:

• A variable called DATETIME must exist in the input data set. For EVENT type tables, DATETIME is the datetime that the event occurred. For INTERVAL type tables, datetime is usually the start of the interval. If you wish to use an alternative datetime variable that is not named DATETIME then you should use the DATETIME= parameter to identify that variable.
  
  
• Additionally, if the table being created is an INTERVAL type table, a variable called DURATION must exist in the input data set. DURATION is the total time (length) of the interval. If you wish to use an alternative duration variable that is not named DURATION then you should use the DURATION= parameter to identify that variable.

Once the source statements have been generated, they must be reviewed before being executed. While GENERATE SOURCE endeavors to be as accurate as possible, it is not infallible in some of the assumptions it makes. Specifically, the generated definitions should be examined to verify (and modify if necessary) the following things:

• On the CREATE TABLE statement, the default age limit for each of the various reduction levels should be reviewed and, if necessary, modified.
  
  
• A CREATE VARIABLE statement is built for each variable that will be in the table. An example of such a statement is as follows:

         CREATE VARIABLE NAME=C EXTNAME=CEXT
           TYPE=NUMERIC LENGTH=8 INTERPRET=COUNT KEPT=YES
           SUBJECT='N/A' LABEL='Int var c'
           DESCRIPTION='Integer variable c'
           FORMAT=BEST12.2
           ISTATS =(
           NOCOUNT SUM AVERAGE NOMAXIMUM NOMINIMUM NOUSS NOCV NORANGE
           NOVARIANCE NOSTD NONMISS
           )
		 DAY =(
           NOCOUNT SUM AVERAGE NOMAXIMUM NOMINIMUM NOUSS NOCV NORANGE
           NOVARIANCE NOSTD NONMISS
           )
		 WEEK =(
           NOCOUNT SUM AVERAGE NOMAXIMUM NOMINIMUM NOUSS NOCV NORANGE
           NOVARIANCE NOSTD NONMISS
           )
		 MONTH =(
           NOCOUNT SUM AVERAGE NOMAXIMUM NOMINIMUM NOUSS NOCV NORANGE
           NOVARIANCE NOSTD NONMISS
           )
		 YEAR =(
           NOCOUNT SUM AVERAGE NOMAXIMUM NOMINIMUM NOUSS NOCV NORANGE
           NOVARIANCE NOSTD NONMISS
           )
         ;

• Also each CREATE VARIABLE statement should be examined to verify that the correct interpretation type is being set by the INTERPRET= parameter.

          if ...                            assume interpretation type 
          ---------------------------       ---------------------------

          the variable format is DATETIMEx  DATETIME

          the variable format is TIMEx      TIME

          the variable format is DATEx      DATE

          the variable format is PERCENTx   PERCENT

          the variable label contains       MINIMUM
          the strings MIN or MINIMUM or
		LWM or LOWEST

          the variable label contains       MAXIMUM
          the strings MAX or MAXIMUM or
		HWM or PEAK or HIGHEST

          the variable label contains       COUNT
          the string COUNT or TOTAL or
		PAGES

          the variable label contains       DATETIME
          the string DATETIME

          the variable label contains       TIME
          the string TIME

          the variable label contains       DATE
          the string DATE

          the variable label contains       PERCENT100
          the string PCT or PERCENT or
		% or UTILIZ or UTILIS

	  the variable label contains       PERCENT
          the string RATIO

  	the variable label contains       GAUGE
          the string CURRENT or SIZE or
		GAUGE

  	the variable label contains       SUM
          the string SUM

          the variable label contains       RATE
          the string PER or RATE or /HR
		or /HOUR or /MIN or /SEC

          the variable label contains       AVERAGE
          the string AVG, AVERAGE, or
          MEAN or AVE

  	the variable label contains       INT
          the string VERSION or NUMBER or
          TYPE or CLASS or PRIORITY or PTY

• Also each CREATE VARIABLE statement should be examined to see if the name of the variable is easy to understand. If not, use a more meaningful variable name.

The syntax of the GENERATE SOURCE statement is as follows:

     GENERATE SOURCE
    < NAME= new-table-name >
      TABLE= < input-table-name | _ALL_ > |
      DATASET= 'input-SAS-data-set-name' |
	INFILE= 'input-file-name'
    < TYPE= INTERVAL | EVENT >
    < INTYPE= DSI >                            see comments below
      ENTRYNAME= 'output-SAS-catalog-entry'
      FILENAME= 'output-physical-file-name'
    < DATETIME= variable-name >
    < DURATION= variable-name >
    < REPLACE | NOREPLACE >
    < NOERROR >
    ;

where the parameters are as follows:

• NAME= table_name

  Names the table to use in the CREATE TABLE NAME= statement that is generated. The table name must be unique within the PDB.

• TABLE= < input-table-name | _ALL_ >

  Names the PDB's IT Service Vision table to use as input for the creation of the dictionary control statements. This table must already exist in the PDB.

• DATASET= 'input-SAS-data-set-name'

  Names the SAS data set to use as input for the creation of the dictionary control statements. This data set must be defined in a SAS library. This parameter must be enclosed in single quotes.

• INFILE= 'input-file-name'

  Names a file (normally a text file) that contains syntax that GENERATE SOURCE has been designed to recognize. This parameter requires that the INTYPE parameter be specified so that GENERATE SOURCE knows what format to expect the text file.

• TYPE= INTERVAL | EVENT

  Only valid when DATASET= or INFILE= are also specified. Sets the type of the table whose source is being generated. The default type is EVENT.

• INTYPE= DSI

  Only valid when INFILE= is also specified. This parameter is used to indicate the source of the file pointed to by the INFILE= statement so that GENERATE SOURCE knows how to process the input file. At the time of writing, DSI is the only INTYPE that can be specified although we expect to add to this in the future.

• ENTRYNAME= 'output-SAS-catalog-entry'

  Names the destination source catalog entry to receive source control statements. This parameter must be enclosed in single quotes.

• FILENAME= 'output-physical-file-name'

  Names the destination file to receive source control statements. This parameter must be enclosed in single quotes.

• DATETIME= varibable-name

  Specifies an alternative variable to represent the default DATETIME variable.

• DURATION= varibable-name

  Specifies an alternative variable to represent the default DURATION variable.

• REPLACE | NOREPLACE

  When REPLACE is used, any existing source in either a catalog or a file will be replaced. When NOREPLACE is use, new source will be appended to the existing source, if there is existing source entry, or will start a new source entry if there is no existing source entry with that name.

• NOERROR

  Indicates that errors caused by this statement should not cause subsequent %CPDDUTL control statements to be skipped.