PK77782: EGL debugger produces EGL0754E with EGL code defining Serial Rec ord with variable length

APAR status

• Closed as program error.

Error description

• Environment: Windows
  
  Product name: IBM? Rational? Business Developer
  
  Program Version    :  7.1.0.1
  Description.
  When running the EGL debugger and stepping through EGL code that
   defines the usage of a Serial Record with variable length it fa
  ils an error is produced if that record consists of Unicode vari
  ables.
  
  The usage of a Serial Record with variable length isn?t possible
   if that record consists of Unicode variables.
  When using char as variable type inside the serial record, the l
  ength needs to be the same amount of bytes (or smaller) then the
   length of the variable.
  
  If you use an amount of bytes that is smaller, the program goes
  into the OnException.
  In System Variables Syslib.currentException the following error
  is shown: (any: string) 'EGL0754E The record length item must co
  ntain a value that splits non-character data at item boundaries.
  EGL0002I The error occurred in the y4fn99 program processing the
   with_Error function.'
  
  The same error will be output from the generated Java produced b
  y the example code segment and this issue is not specific to deb
  ugger functionality it relates to language restriction developme
  nt have confirmed via: http://publib.boulder.ibm.com/infocenter/
  rbdhelp/v7r5m0/index.jsp?topic=/com.ibm.egl.lr.doc/topics/regl_d
  ata_file_length_item.html.
  
  However the actual generated COBOL code will include the variabl
  e length record which the COBOL implementation supports so the u
  ser finds he has to run the debugger with only a fixed length re
  cord and alter their code to generate as a variable length recor
  d.
  
  The user requests that the debugger should work with any stateme
  nts which are supported by the language which is being generated
  .
  
  Example code:
  record utfile1 type SerialRecord  {fileName = 'utfile1', lengthI
  tem = ut_Lgh1}
  // ---------------------
    01 txt   char(100);
  end
  
  function correct()
    wdybok.ut_Lgh1 = 4;
    set utfile1 empty;
    utfile1.txt = 'Gery';
  
    try
     add utfile1;
    onException
     SysLib.writeStdout('error utfile1') ;
    end
  
   end
  
  When using Unicode as variable type, you can only use the same a
  mount of bytes for the variable length.
  
  record utfile2 type SerialRecord  {fileName = 'utfile2', lengthI
  tem = ut_Lgh2}
  // ---------------------
    01 txt   unicode(50);
  end
  
  function with_Error();
    wdybok.ut_Lgh2 = 4;
    set utfile2 empty;
    utfile2.txt = 'Gery';
  
    try
     add utfile2;
    onException
     SysLib.writeStdout('error utfile2') ;
    end
  
   end
  

Local fix

Problem summary

• EGL was throwing an exception during execution of an add
  statement, if it used a variable-length record, and the
  record's lengthItem indicated that the data ended in the
  middle of a field whose type was not char.  This happened in
  the EGL debugger and generated Java programs.
  

Problem conclusion

• The operation is now allowed when the field is a char, hex,
  or unicode (but a unicode character may not be cut in half
  or the exception will still be thrown).
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  RATL BUS DEV

• Fixed component ID

  5724S5000

Applicable component levels

• R710 PSN

     UP