GnuCOBOL / Bugs / #1244 GCSORT support of DFSORT parameters SYMBOLS

Sauro Menna - 2026-08-21

Hi,
I've analyzed the issue regarding SYMBOLS.
Indeed, the SORT function must handle the SYMBOLS.

For GCSort, I thought I'd organize the handling this way:
1) Define an environment variable named: SYMNAMES
2) Assign the name of a file containing the symbolic variable information to this variable

Example: SYMNAMES=..\files\symnames.txt

The file ..\files\symnames.txt will contain:
<variable name="">,value(s)</variable>

An example of the file format is as follows:
ID,1,8,ZD
Name,9,15,CH
If the first character is a ‘*’, the line indicates a comment.

During startup, if GCSort finds the SYMNAMES environment variable defined, it retrieves the file name.
If the file exists and its format conforms to the expected structure, it loads a table of symbols that
GCSort will use to replace variables with the values extracted from the file.
The substitution will occur on all lines sent to GCSort, before actual execution.

I’ll need a few days to implement the solution.
I’ll send an update as soon as I finish development and testing.
Okay?
Best regards.

Sauro

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Simon Sobisch - 2026-08-21
  
  Sounds like a good plan - but I suggest to use libcob's file resolution instead of building that manually (this way DD_SYMNAMES and SYMNAMES env vars work, as well as using COB_FILE_PATH). The easiest way is possible calling into cob_open for input - and either read_next through the file (then it exists somewhere following gc rules) or don't do the symnames setup.
  
  Last edit: Simon Sobisch 2026-08-21
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
  - Sauro Menna - 2026-08-23
    
    Okay,
    thanks for the suggestion.
    The feature will be implemented using libcob.
    The question is which file type to use (Line Sequential or Sequential).
    If Line Sequential is used, we might run into the problem previously identified in the GCSort Take file, namely a NULL character in the file.
    For now, I’ll use Line Sequential, so I’ll need to generate the SYMNAMES files with that structure (without NULL characters).
    
    If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
    - Ralph Linkletter - 2026-08-23
      
      Are you planning to use the takefile for specifying the SYMBOLS ?
      
      SYMBOLS ORDERNUM,1,6,CH USE C:\WINZOS\DATA\SORTIN03.DAT RECORD F,00400 ORG SQ USE C:\WINZOS\DATA\SORTIN02.DAT RECORD F,00400 ORG SQ USE C:\WINZOS\DATA\SORTIN01.DAT RECORD F,00400 ORG SQ GIVE C:\WINZOS\ZOSICF\S2121566.DAT RECORD F,00400 ORG SQ OPTION MODS E15=(TESTAE15) SORT FIELDS=(ORDERNUM,A) OMIT COND=(3,1,CH,EQ,C'1')
      
      That actually seems as if would fit in easily - would it not ?
      Are there other operands that precede the USE clause(s) ?
      
      Please advise
      
      If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
      - Sauro Menna - 7 days ago
        
        Hi,
        The implementation in GCSort will be the same as in DFSORT, that is, using an external file, as in JCL.
        In accordance with the DFSORT standard, an environment variable named DD_SYMNAMES, dd_SYMNAMES, or SYMNAMES must be created, specifying the physical file.
        According to IBM and Microfocus documentation, the dataset or file assigned to DD SYMNAMES must have a fixed-length record format (RECFM=F or RECFM=FB) and a record length of 80 bytes (LRECL=80).
        I will adopt this definition in GCSort.
        Example:
        
        //SYMNAMES DD * ID_DIP,1,5,CH LAST_NAME,6,15,CH SALARY,21,4,BI
        
        or with the SKIP parameter
        
        //SYMNAMES DD * ID_DIP,1,5,CH * Starts at column 1. The next free byte is 6. ,*,SKIP,50 * Skips 50 bytes. The pointer moves to column 56. LAST NAME,*,15,CH * Starts automatically at 56. The next free byte is 71. ,*,SKIP,10 * Skips 10 bytes. The pointer moves to column 81. SALARY,*,4,BI * Starts automatically at 81.
        
        The Microfocus Sort utility allows you to specify symbols in the TAKE file, but they must be enclosed within a block.
        Example:
        
        * -------------------------------- ------------------ * Section 1: Symbol Dictionary Definition * -------------------------------------------------- SYMNAMES ID_DIP,1,5,CH * Column 1 for 5 bytes ,*,SKIP,50 * SKIP 50 bytes (columns 6–55) LAST_NAME,*,15,CH * Column 56 for 15 bytes ,*,SKIP,10 * SKIP 10 bytes (columns 71–80) SALARY,*,4,BI * Column 81 for 4 bytes (binary) SALARY_THRESHOLD,+50000 END * Ends the SYMNAMES block * -------------------------------------------------- * Section 2: Data Processing Commands (Logic) * -------------------------------------------------- INCLUDE COND=(SALARY,GT,SALARY_THRESHOLD) SORT FIELDS=(LAST_NAME,A)
        
        I think the idea of including the symbol definitions in the TAKE file is a good one.
        I’ll implement this using both an environment variable with an external file and within the TAKE file itself.
        
        Thank you.
        Sauro
        
        👍
        1
        
        Last edit: Simon Sobisch 7 days ago
        
        If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
        
        Ralph Linkletter - 6 days ago
        
        I understand the SYMBOLS integration.
        
        Hey while you are in the source code can you add a counter to the audit file.
        Records omitted or included.
        This is what I produced from my DFSORT emulation.
        
        Thank you for the feedback !
        
        Bruce
        
        +++ DFSORT Processed RECORDS IN=: 262 +++ DFSORT Processed RECORDS OUT=: 74 +++ DFSORT Processed OMIT/INCLUDE=: 188 <=========* ** WINZOS JOB LOG Copyright 2010-2025 Metamorphous Company LLC ** JOB: AY2SORT JOB NUMBER: 0409 DATE: 08/24/26 TIME: 11:50:37 1 //AY2SORT JOB 'WINZOS-SORT',CLASS=A,MSGCLASS=A 2 //STEP01 EXEC PGM=IEFBR14 3 //SYSUT1 DD DSN=TESTDATA.WINZOS.SORTED.DATA,DISP=(MOD,DELETE,DELETE) 4 /* 5 //STEP02 EXEC PGM=SORT 6 //SORTIN DD DSN=TESTDATA.UNSORTED.ASCII.DATA,DISP=SHR 7 *** DD DSN=TESTDATA.UNSORT.ASCII.DATA2,DISP=SHR 8 *** DD DSN=TESTDATA.ASCII.DATA,DISP=SHR 9 //SORTOUT DD DSN=TESTDATA.WINZOS.SORTED.DATA,DISP=(NEW,CATLG,DELETE), 10 // DCB=(LRECL=400,RECFM=FB,BLKSIZE=0) 11 //SYSOUT DD SYSOUT=* 12 //SYSIN DD * 17 *** INCLUDE COND=(3,1,CH,EQ,C'1') 18 *** SORT FIELDS=(1,6,Y2C,A) 19 *** SORT FIELDS=(393,4,Y2P,A) 20 *** SORT FIELDS=(393,4,BI,A) 21 *** SORT FIELDS=(1,6,CH,D,7,12,CH,A) 22 // **** JOB STATISTICS **** 11:50:37 STEP01 - STEP STARTED TESTDATA.WINZOS.SORTED.DATA DELETED 11:50:37 STEP01 - STEP WAS EXECUTED RETURN-CODE=0000 11:50:37 STEP02 - STEP STARTED +++ DFSORT Processed RECORDS IN=: 262 +++ DFSORT Processed RECORDS OUT=: 74 *+++ DFSORT Processed OMIT/INCLUDE=: 188* TESTDATA.UNSORTED.ASCII.DATA KEPT TESTDATA.WINZOS.SORTED.DATA CATALOGED C:\WINZOS\ZOSICF\04090200.003 SYSOUT &&TEMP.SY020004.409 DELETED 11:50:37 STEP02 - STEP WAS EXECUTED RETURN-CODE=0000 11:50:37 AY2SORT (0409) - JOB ENDED
        
        👍
        1
        
        Last edit: Simon Sobisch 6 days ago
        
        If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Simon Sobisch - 2026-08-21

labels: --> GCSORT

status: open --> accepted
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Ralph Linkletter - 2026-08-21

I would think that the implementation would follow SYSIN
From a zOS perspective there is a SYSIN DD and a SYMNAMES DD

~~~
//STEP02 EXEC PGM=SORT
//SORTIN DD DSN=TESTDATA.UNSORTED.ASCII.DATA,DISP=SHR
// DD DSN=TESTDATA.UNSORT.ASCII.DATA2,DISP=SHR
// DD DSN=TESTDATA.ASCII.DATA,DISP=SHR
//SORTOUT DD DSN=TESTDATA.WINZOS.SORTED.DATA,DISP=(NEW,CATLG,DELETE),
// DCB=(LRECL=400,RECFM=FB,BLKSIZE=0)
//SYSOUT DD SYSOUT=*
//SYSIN DD DSN=TSUBLUH.SORTCNTL.PDSIN,DISP=SHR

//STEP02 EXEC PGM=SORT
//SORTIN DD DSN=TESTDATA.ASCII.DATA,DISP=SHR
// DD DSN=TESTDATA.UNSORT.ASCII.DATA2,DISP=SHR
// DD DSN=TESTDATA.UNSORTED.ASCII.DATA,DISP=SHR
// DD DSN=TESTDATA.ASCII.DATA,DISP=SHR
//SORTOUT DD DSN=TESTDATA.WINZOS.SORTED.DATA,DISP=(NEW,CATLG,DELETE),
// DCB=(LRECL=400,RECFM=FB,BLKSIZE=0)
//SORTCNTL DD *
OPTION MODS E15=(TESTAE15)
SORT FIELDS=(1,6,CH,A)
OMIT COND=(3,1,CH,EQ,C'1')
/

The SYSIN DD dataset name is a defacto zOS best practice
As opposed to SYSIN DD *

The SYMNAMES parameters should parallel the SYSIN paradigm

However SYMNAMES is implemented I have the need to extract it from JCL and feed GCSORT - Akin to SYSIN

Last edit: Ralph Linkletter 2026-08-21

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Sauro Menna - 4 days ago

Hi,
I've released a new version of GCSort, ver. 1.05.00, with a number of updates.
Here are the details:

- Added support for reading the TAKE file using libcob. - SYMNAMES symbol integration [1] SYMNAMES with definitions in an external file using libcob [2] SYMNAMES definitions in the TAKE file - A new counter has been added to the print statistics for OMIT/INCLUDE records - Updated the configuration files: gctestsetup.cfg and gctestcmd.cfg for test cases related to SYMNAMES

Best regards.
Sauro
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
- Ralph Linkletter - 4 days ago
  
  Can you document how it was implemented on your side please.
  SYMORG or SYMNAMES ?
  If in the TAKEFILE what syntax is required in supporting symbols
  The new record counter is impleneted with ?
  Please advise,
  
  Thanks
  Bruce
  
  If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
  - Sauro Menna - 4 days ago
    
    Hi,
    Attached are two examples of how to use SYMNAME.
    1) SYMNAMES within the TAKE file. In this case, the contents of the SYMNAMES definition are enclosed between the SYMNAMES (start) and END (end) lines. The definition of the SYMNAMES and SYMORG environment variables must be omitted.
    Within the TAKE file, before defining the SORT instructions, you can insert symbols enclosed between two lines that define the start and end of the SYMNAMES definition.
    Start: SYMNAMES
    End: END
    
    2) SYMNAMES defined in an external file. In this case, you must set the SYMNAMES environment variable to the pathname of the definition file. SYMORG is required only for SYMORG=SQ, and the record length is fixed at 80 characters.
    The requirements for the definition are identical to those for the TAKE file.
    
    The definition of the SYMNAMES fields is as follows:
    SYMNAMES format:
    
    [1] <var name>,value [2] <var name>,pos,len,type [3] <var name>,*,len,type [4] SKIP,value
    
    In the SYMNAMES file, you can assign a descriptive name to record positions, constants, or keywords.
    The asterisk (*) indicates “continue from where the previous field ended.”
    
    I have included some additional details in the document located in gcsort/doc (GCSORT_Manual.pdf).
    
    The new record counter dedicated to OMIT/INCLUDE is displayed only if the SORT commands contain the OMIT or INCLUDE construct and the counter takes on a value greater than zero.
    Example:
    ==============================================================
    
    Total Records Number..............: 786 Total Records Write Sort..........: 0 Total Records Write Output........: 222 Total Records Omit/Include........: 564
    
    ==============================================================
    
    symnames_example2.txt
    
    If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
    - Sauro Menna - 4 days ago
      
      first example
      
      symnames_example1.txt
      
      If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Sauro Menna - 4 days ago

Hi,
Attached is a ZIP file with two examples in Windows environment.
readme.txt contains the details.

1) Example - SYMNAMES definition in the take file
gcsort TAKE take_symb.prm
2) Example - SYMNAMES definition in an external file
set SYMNAMES=fsymb.txt
gcsort TAKE take_symb_file.prm

Sauro

symnames_example.zip

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

GCSORT support of DFSORT parameters SYMBOLS

A free COBOL compiler

Group

Searches

Help

#1244 GCSORT support of DFSORT parameters SYMBOLS

Discussion