Columns

Columns is used to limit the number of columns associated with a table:
* Loaded in the class [F] when reading the data by a For, Readlock or Read instruction.
* Updated from the class [F] when writing the data by a Rewrite or RewriteByKey instruction.

Syntax

Columns CLASS (FIELD_LIST)
Columns CLASS (FIELD_LIST) Extended
Columns CLASS

• CLASS is a class associated with the table opened by File, or a join opened by Link.
• FIELD_LIST is a list of columns in the table or join, separated by commas. The columns can be prefixed by the table abbreviation. This is especially useful when used in conjunction with a Link.

Examples

   # Read then rewrite limited to 2 fields, code and file name
   Local File BPARTNER [BPR]
   # Position of the filter on the code and client name
   Columns [BPR](BPRNUM,BPRNAM)
   For [BPR]
       ...
       Rewrite [BPR]
   Next
   # Filter deletion: all the columns are now available again
   Columns [BPR]

  # Column filter on the class derived from the link
  Local File ORDERS [ORD]
  Local File ITMMASTER [ITM]
  Link [ORD] with [ITM]ITM0=[F:ORD]ITMREF as [ORI]
  # Position of the filter on the product ref., the product dest., the order no.
  Columns [ORI]([ITM]ITMREF,[ITM]ITMDES1,[ORD]WIPNUM)
  For [ORI]
       ...
  Next
  # Filter deletion: all the columns are now available again
  Columns [ORI]

  # Column filter on the class that applies also on Read instruction
  Local File ORDERS [ORD]
  Columns [ORD] (ORDNUM,CUSTOMER) Extended
  Read [ORD]ORD0=MY_ORDER
  If [ORD]CUSTOMER=MY_CUSTOMER
    ...
  Endif

  # Column filters that excludes some columns
  # - those having some data types (Clob, Blob, strings with a max length over 50 characters)
  # - for columns having a dimension greater than 1, we will only include the 2 first indexes
  Local File ORDERS [ORD]
  # Let's first build the INCLUDE_LIST array by using the meta data stored in [G:ORD] class
  Local Char INCLUDE_LIST(250)
  Local Integer I,J,K
  INCLUDE_LIST=""
  For I=1 to dim([G:ORD]nbzon-1
    J=evalue("type([G:ORD]adxfname("+num$(I)+"))"
    K=evalue("dim([G:ORD]adxfname("+num$(I)+"))"
    If (J<=60 or J>265) and (J<>522) and (J<>523)
      INCLUDE_LIST+=string$(INCLUDE_LIST<>"",",")+[G:ORD]adxfname(I)+string$(K>2,"(0..1)")
    Endif
  Next I
  # Now the Columns can be done by evaluating a string constant that contains the list
  Columns [ORD] (=evalue('"'+INCLUDE_LIST+'"'))

Description and comments

• By default, Read and For instructions perform select * SQL sentences on the database. This may be very costly if a large number of columns are present on the table. Columns is used to limit the access to the columns; in that case, the select sentence executed by the engine will send the list of columns given by the Columns instruction.
• The use of Columns is very cost effective when a large set of data is extracted from the database (usually with the For instruction). When a unique record is accessed by Read or Readlock, the cost is lower, and thus Columns does not apply on Read or Readlock except if the Extended keyword is used.
• Columns without a list of fields restore the default behavior for the future requests done on the table.
• At update time, the Columns instruction is not taken into account by the Write instruction that still manages class[F] as a whole, but only by Rewrite and RewriteByKey.
• The class [F] variables excluded by the Columns instructions are still present, but they are no longer loaded or used for update by the corresponding instructions.
• A Columns instruction replaces a previous Columns instruction done on the same class.
• Elements in a column can be evaluated specifically with the syntax =evalue(EXPRESSION) where EXPRESSION is a string expression.
• When elements have a dimension, it is possible to limit the number of indexes considered by the syntax FIELD(INDEX1..INDEX2) where FIELD is a column of the table and INDEX1 and INDEX2 are constants giving the ranges of index.
• The column list for a table opened by Local File is managed at the right nesting level in the script. Thus, if a Columns instruction on a table is declared with Local File keywords, and a nested script executes another Local File keywords with a new Columns instruction, then the the list of table columns involved in the upper level Columns instruction remains unchanged after the execution of the inner level Columns instruction even if this one concerns another list of columns on the same table. This last point can be illustrated by the following example:

   # Read data in a loop with a filter on properties
   Local File BPARTNER [BPR]
   # Position of the filter on the code and client name
   Columns [BPR](BPRNUM,BPRNAM)
   For [BPR]
       ...
   Next

   # Call another subprogram
   Call MY_SUB

   # Read again the data
   # The data read is still restricted here to the BPRNUM and BPRNAM columns
   For [BPR] Where...
       ...
   Next
End

Subprog MY_SUB

   # Read data in the same table, but restricted on other properties
   Local File BPARTNER [BPR]
   # Position of the filter on the code and client name
   Columns [BPR](FCY,CRY,CRN)
   For [BPR]
       ...
   Next
End

Errors

See also

File, Link, For, Read, Rewrite, RewriteByKey.