Prallel Routine

How to create shared C code object and use them in datastage parallel routine?

Hello all, in this post i will give you information on datastage parallel routines and shared C code objects, and how can we create shared C code objects and use them in parallel routines.
Datastage Parallel Routine:
A parallel routine provides you feature to use external functionality written in C code to use in Datastage.
E.g.  Datastage does not provide regular expression functionality. So we can created shared object of regular expression functionality in C and used it in Datastage.
Steps:
1.       Create required function in C.

Here is a simple C function to add two numbers.
No need of Main.

#include
 
int sum_p(int a , int b)
{
c=a+b;
return c;
}

Suppose the name of above program is sum_pk.c

2.       Create shared object /library of the code.

Position Independent Object:
g++34 -fpic -c sum_pk.c

g++: GNU compiler available in Unix. g++34 is version of g++ available on our server.
-c : compiles code and creates object of file
-fpic: creates object with position independent code which is required for shared object/library

Object file with extension  .o will be created as sum_pk.o

a)      Shared Object:
Shared object is created from position independent object file created above.
g++34 -shared -o sum_pk.so  sum_pk.o

sum_pk.so is the shared object file created from sum_pk.o

b)      Shared library:
Shared library is also created from position independent object file created above.
g++34 -shared -o libsum_pk.so sum_pk.o

libsum_pk.so is the shared library file created from sum_pk.o

Shared library Vs Shared Object:

Shared Library	Shared object
A shared library file is linked to job at runtime and must be available at runtime.	A shared object file is linked to job at compile time.
Shared library name should start with “lib” and should have “.so” as extension E.g. libsum_pk.so	No such constraint on shared object.
Shared library should be present in predefined library paths. E.g. /opt/IBM/InformationServer/ASBNode/lib/cpp/ is the library path in our datastage installation	No such constraint on shared object.

3.       Creating a parallel routine in Datastage:

• File>New>Routines>Parallel Routine
• Fill all the required values as:

Routine Name:  Any name with just alphanumeric characters only. No underscore as well.
External subroutine name: Name of the C function which we want to invoke
Type: External Function
Object Type: Library if you are using shared library or Object if you are using shared object.
Return Type: Return type of the C function
Library path: Library name with complete path
If shared library the path should be
                            /opt/IBM/InformationServer/ASBNode/lib/cpp/

Go to Arguments tab and enter details about the input arguments that the C function/ parallel routine will consume.
Save your routine to required folder.

Usage:

You need a transformer stage to use parallel routine external function in your job.
In Derivation pane of any port in transformer stage:
Right Click > DS_Routines> will call your riutine
E.g. SumRoutinePK(%a%, %b%)
Enter the required input parameters and it will return the required result.
sumRoutinePK(DSLink2.F1, DSLink2.F2)