Database Output from CL

I know that IBM added a lot of new programming features to CL in the past two releases. However, one of the features I’ve been hoping for–the ability to do output operations to database files–doesn’t appear to be among them. Please tell me that I’m wrong and that there really is a way to update a database record within a CL program.

–David

You’re right in thinking that IBM has not added CL commands to write, delete, and update database records, David, but that does not mean that there is no way to update a record in a database file.

Let me back up and say that CL has always had a limited ability to write to the database. Every command that includes an OUTPUT parameter and allows the *OUTFILE option writes to a database file. This uses Display Object Description (DSPOBJD) command:

DSPOBJD OBJ(THOLT/S*)  OBJTYPE(*PGM) +
           OUTPUT(*OUTFILE) OUTFILE(QTEMP/DSPOBJD)

Even if a command does not allow output to an outfile, you may still be able to route printed data into a database file. The following sequence of commands places the output of the Display Library List (DSPLIBL) command into program-described file LIBL in QTEMP.

OVRDBF  FILE(QPRTLIBL) TOFILE(QTEMP/LIBL) LVLCHK(*NO)
CRTPF   FILE(QTEMP/LIBL) RCDLEN(132)
DSPLIBL OUTPUT(*PRINT)

You can also use Copy File (CPYF) and Copy from Query File (CPYFRMQRYF) commands to add to or replace the contents of a database file. (Did you see the tip called Where’s the Other MBROPT Option? I got some good responses to that one.)

So CL has always had some ability to update and add to database files, but this does not address your question, which is how to update a single record in a database file.

I suggest you look at Qshell’s db2 utility, which allows you to run SQL commands. When I first wrote about the db2 utility, IBM did not officially support it. As far as I know, that is still the case. My reasoning for that is that db2 is not listed among the utilities in the V5R4 Qshell reference.

The db2 utility cannot process a cursor, so there is no way for you to use Receive File (RCVF) to read a record, then use db2 to update same the record, as you can do using SQL in a high-level language like RPG or COBOL. You will have to put enough expressions in the WHERE clause to isolate the record you want to update. If you can’t isolate the record, db2 won’t do the job for you and you’ll have to resort to another language.

In the following example that I quickly threw together, the program attempts to read a record whose ID field has the value INVOICE. If that record is found, the program adds one to the batch sequence number and updates the record. If the record is not found, the program creates a record with a key value of INVOICE.

pgm                                                             
                                                                
  dcl  &Command    *char    256                                 
  dcl  &BatchNbrA  *char      5                                 
  dcl  &Lib        *char     10                                 
                                                                
  dclf BatchSeq                                                 
  
  /* Don't present a Qshell terminal session. */ 
  addenvvar  envvar(QIBM_QSH_CMD_OUTPUT)     value('NONE')      
  monmsg     cpfa980                                            
                                               
  /* Position to the INVOICE record in BatchSeq. */ 
  ovrdbf  file(BatchSeq) position(*keyae 1 rec 'INVOICE')       
  rcvf                                                          
  monmsg  cpf4137                                               
                                                  
  /* The db2 command needs to know which library the file is in. */
  rtvobjd   obj(BatchSeq) objtype(*file) rtnlib(&lib)           
                                                                
  /* Add or update the INVOICE record with the last batch number. */
  if (&ID *eq 'INVOICE') do                                     
     chgvar &BatchNbr   (&BatchNbr + 1)                         
     chgvar &BatchNbrA  &BatchNbr                               
     chgvar &Command    ('update' *bcat &lib *tcat '.BatchSeq +
                             set BatchNbr=' *cat &BatchNbrA *cat +         
                         ' where ID=''INVOICE''')                          
  enddo                                                                    
  else do                                                                  
     chgvar &BatchNbr   1                                                  
     chgvar &BatchNbrA  &BatchNbr                                          
     chgvar &Command    ('insert into' *bcat &lib *tcat '.BatchSeq +       
                            values(''INVOICE'',' *cat &BatchNbrA *cat ')') 
  enddo                                                                    
                                                                           
  chgvar &Command ('db2 "' *cat &Command *tcat '"')                        
  qsh    cmd(&Command)

The generated db2 commands look something like this:

QSH CMD('db2 "update MYLIB.BatchSeq set BatchNbr=00003 
		where ID=''INVOICE''"')

QSH CMD('db2 "insert into MYLIB.BatchSeq 
		values(''INVOICE'',00001)"')

Before you decide whether or not to use the db2 utility, you need to be aware of a problem. This command always returns a zero status, whether the command executes properly or not. Such behavior does not do wonders for error-checking, and so far I have not found a good solution to it.

–Ted

RELATED STORIES

Where’s the Other MBROPT Option?

Updating Through an SQL Cursor

Access the Database from Qshell