Re: Terminating a C proram

["Pantzopoulos, Mike" <mikepantzopoulos@xxxxxxxx>]

Title: Re: Terminating a C proram

Thanks to all who helped with a response to the above subject. My summary of the answers and further research so far is:

1. EPM does not exist and  is of no concern to the current C compiler and run-time.. This means that the Jennifer Hamilton book is well out of date? I couldn't find any mention of Activation Groups in the index, and the first (and only!) publishing date was October 1992!!. For $200 Australian I would expect that the publishers keep the book up to date. I feel a little skinned

       

2. All I am trying to achieve is a consistency with RPG- to retain the C program and its lower level RPG invocations in memory until the business function has ended, in which case I will pass through a request code that results in the lower level programs setting on *INLR and then results in the C program ending as well.


3. I looked up the ILE C Programmers Guide(appendixH) where it states.....keeping an activation group active after the program has exited means that all the storage associated with that program is still allocated and in a 'last-used' state. When the program is called again, initialization, as defined by the language, is not done. All variables have the last value assigned to them in the previous run of he program. In addition all of the settings in the ILE C run-time are in 'last-used' state such as signal(),strtok() etc......." .(reference H1.1).



..............no mention of the static attribute............

In the same section they provide some example code and so I  created two C programs (copied from H1-1 in the users guide), one called TESTRTRN (with a return;) and the other TESTEXIT(with the return replaced by an exit() with test harnesses as follows:




#include <stdio.h>                                                     
        int invocation_count=0;                                            
        int main(void)                                                     
        {                                                                  
        invocation_count++; /* increment counter */                        
        printf("Exit called: ");  ..........for the TESTRTRN program......printf("Return called: ");                                         

        printf("%d\n", invocation_count);                                  
        getchar(); /*  this holds the screen open */                       
exit();..........for the TESTRTRN program........... return;                                                           

}

I set up a 3 tier calling structure with CL programs:
                                TEST 1                  TEST 2 

                                Activation              Activation
TProgram                        Group                   Group
TESTBOTHCL (calls)      *NEW                    *NEW

TESTEXITCL (calls)      *CALLER         *CALLER
                TESTEXIT        *NEW                    named
                TESTEXIT           "                       "
                TESTEXIT           "                       "
                TESTEXIT           "                       "
                TESTEXIT           "                       "




TESTRTRNCL (calls)      *CALLER         *CALLER
        TESTRTRN        *NEW                    named
        TESTRTRN           "                       "
        TESTRTRN           "                       "                   
        TESTRTRN           "                       "

TESTRTRN           "                       "

               
TEST 2 worked as I wanted. The value of I was retained between invocations for the TESTRTRN program The TESTEXIT program also worked correctly with I being output as 1 on each call.

TEST 1 printed I=1 each time it was called. Correct for the TESTEXIT program but not for the TESTRTRN program.

I thought perhaps that *NEW and named would work in the same way but unfortunately no! I'm still unsure about the difference. More manual bashing I think.

This TEST 2scenario seems like it will solve my problem, as the return corresponds to the RPG return, as does the exit() to SETON *INLR as long as the activation group is named.

 I thought this issue might of interest to others approaching C from a mainly RPG frame of reference. Any further comments would still be appreciated

Mike Pantzopoulos