SEARCH and SEARCH ALL in COBOL with examples

SEARCH is a serial search.It can be used for both sorted and unsorted data inside the table.The table needs to be defined with an index .When we use the search function, the function starts with the current setting of the index and it goes on until to the end of the table.The starting value of the index is left upto the choice of the programmer. We can set the value of the index value just before the search starts using the SET keyword.

IDENTIFICATION DIVISION.
PROGRAM-ID. HELLO.

DATA DIVISION.
WORKING-STORAGE SECTION.
   01 WS-TABLE.
          05 WS-TEST PIC X(1) OCCURS 10 TIMES INDEXED BY I.

   01 WS-SRCH-STRNG PIC A(1) VALUE 'R'.

PROCEDURE DIVISION.
   MOVE 'ABCDEFRJKL' TO WS-TABLE.
   SET I TO 1.
   SEARCH WS-TEST
     AT END DISPLAY   'R NOT FOUND IN TABLE'
     WHEN WS-TEST(I)=WS-SRCH
     DISPLAY    'WE FOUND YOU IN THE TABLE'
   END-SEARCH.  

STOP RUN.

This linear Search will iterate through the entire table until the search-string is found or the end of table is reached.
In this case, since our search string 'R' is there in the table, the SYSOUT will display 'WE Found.. '.

SEARCH ALL is a binary search and the table must be sorted .That is the the prerequisite to use search all. The outcome of SEARCH ALL is either a yes or a no. Hence we can not code multiple search conditions when using SEARCH ALL function.
Contrary to search function, there is no need to initialize the index .

Check the following code:

IDENTIFICATION DIVISION.
PROGRAM-ID. HELLO.

DATA DIVISION.
   WORKING-STORAGE SECTION.
   01 WS-TABLE.
      05 WS-BIN-SRCH OCCURS 10 TIMES ASCENDING KEY IS WS-NUM INDEXED BY I.
      10 WS-NUM  PIC 9(2).
      10 WS-DEP   PIC A(3).

PROCEDURE DIVISION.
   MOVE '12ABC56DEF34GHI78JKL93MNO11PQR' TO WS-TABLE.
   SEARCH ALL WS-BIN-SRCH
     AT END DISPLAY 'BIN-SRCH NOT FOUND'
     WHEN WS-NUM (I)=93
     DISPLAY 'BIN-SRCH FOUND '
     DISPLAY WS-NUM(I)
     DISPLAY WS-DEP(I)

Just a point to note:  The table is searched in ASCENDING ORDER. If you see, the table is defined with ASCENDING key clause. In case we want, we can always define the table in DESCENDING KEY clause.
To use SEARCH ALL,a field must be in ASCENDING or DESCENDING key clause. This Identifier must be an entry within the table

When to use SEARCH and SEARCH ALL Function:

· For a table with less than 50 entries, go for SEARCH (Sequential Search)

· For a table with more than 50 entries go for SEARCH ALL (Binary Search)

Difference between PDS and PDSE - Brief discussion

We all know that a PDS (partioned data set), commonly referred to as Library in Z/OS world is a collection of several data sets which we call as members. A PDS internally contains a directory which keeps track of the member names.
A PDS is created with a Data Set Organization of PO , (DSORG=PO), which stands for partitioned organization.

Creating PDS using 3.2 option
Directory blocks  . . 20
Data set name type  : PDS
Using JCL:
SPACE=(TRK,(50,10),20)

However a PDS have certain disadvantages:
1. When we delete a member from a PDS, the space remains unused. We need to compress the PDS using 'Z' at the command line , or we can use IBM utility  IEBCOPY to reclaim the unused space.

2. Also as we know, a PDS internally contains a directory. As the size of the PDS grows, ie, as we add more and more members into the pds, the directory size gets near to the threshold (This Directory size we give when we define a PDS), after which we can not add any more members in the PDS.
Then we need to copy all the members into a new PDS with increased Directory size.

PDSE ( partitioned data set extended ):  Exactly same as PDS in many respects.
However , PDSE  data sets can be  stored only on DASD, not on tape. Interesting thing is the directory can expand automatically  as needed. Additionally it has an index which helps to locate the members inside the PDSE faster . SPACE from deleted members are automatically reused in PDSE .

PDSE files have DSORG=PO and DSTYPE=LIBRARY.
JCL to create PDS

//ALLOC    EXEC  PGM=IDCAMS
//SYSPRINT DD    SYSOUT=A
//SYSIN    DD    *
    ALLOC -
    DSNAME(TEST.PDSE1.EXAMPLE1) -
    NEW -
    STORCLAS(RM06) -
    MGMTCLAS(RM06) -
    DSNTYPE(LIBRARY)