Dynamic calls to ENTRY points in COBOL

rockish · Posted: Wed Sep 28, 2011 8:35 am

Hi,

I am trying to make a dynamic call from ProgramA to EntryB. EntryB is an ENTRY point defined in the programB as ENTRY EntryB. When I try to run the program, the job fails with the abend code S806 & I get the following error message.

enrico-sorichetti · Posted: Wed Sep 28, 2011 9:41 am

the first quick and dirty rule is that the entry point must be the load module name
that' s what You see
just think about it, how in heaven is <cobol> supposed to determine the load module name...
plainly and simply from the name of the entry point called...

quoting the manual

Bill Woodger · Posted: Wed Sep 28, 2011 10:52 am

Seperate ENTRY points in Cobol have been out of favour with good programming practice for some time.

Do you have a really specific task which has to be accomplished with an ENTRY point?

If so, how come you don't know how to use it, even though, as enrico points out, the manual covers the topic?

If you don't have something specific, I'd suggest not using an ENTRY point.

If you have a module which provides multiple functions, requiring distinct parts of the program to be executed depending on the call yet sharing the same DATA DIVISION, there is absolutely no problem in handling this with a flag/indicator/parameter whatever you want to call it. Test at the "top" of the program, send the stuff on their seperate routes via performs. All return on completion to a single point, where you might find other things to do but where you at least GOBACK/RETURN.

rockish · Posted: Wed Sep 28, 2011 6:44 pm

Thanks very much for your inputs. I do not have any specific functionality to perform using this. or rather I am not implementing a new functionality at all. I am currently working on an maintenance project which involves converting all the static calls in the current application to dynamic calls. As I was involved in the analysis for the same, I encountered this scenario and was wondering how can I achieve this.

Robert Sample · Posted: Wed Sep 28, 2011 7:02 pm

Do NOT convert the static calls to dynamic calls. As you have already discovered, there will be side effects, not to mention you will have to make code changes -- such as the CANCEL statement required as per enrico's post -- and the future maintenance will be more complicated.

rockish · Posted: Wed Sep 28, 2011 7:08 pm

Thanks Robert for your response. I was always under the impression that converting static to dynamic calls can minimize the maintenance involved because of making every module stand-alone. Though the modules involving ENTRY points is a new lesson for me, I am attracted towards your words that states that the future maintenance will be more complicated because of this. I am just curious to know more disadvantages of converting the static calls to dynamic calls in terms of the mainenance issues. Is there any article that you can guide me to in this regard ?

Bill Woodger · Posted: Wed Sep 28, 2011 7:21 pm

The best thing would be to get rid of all the entry-points, re-design, re-write, change all the calls to the single main entry (program-id). Then you can proceed with the original plan.

Of course, the number of modules and effort involved may preclude this. Then you may end up with a mix of static and dynamic, so you always have to redo the all the loads involved when the module changes.

There may be no clearcut answer to which leads to less maintenance for a given site, because it depends how all the stuff hangs together already. If you have only five load modules representing your entire batch run, yet each contains several hundred modules, static might suit better. If all are basically stand-alone, just using modules for general-purpose stuff, then dynamic might suit better.

If you have some dodgy old code with entry points in....?

enrico-sorichetti · Posted: Wed Sep 28, 2011 7:26 pm

my extra .02 €
if You have to go for full flexibility split everything...
one function/subroutine <==> one entry point <==> one load module
it might avoid fights on how to code the function vectors

Robert Sample · Posted: Wed Sep 28, 2011 7:34 pm

The general statement is that dynamic calls may reduce future maintenance, but it depends upon the specifics. The specifics of your situation indicate that maintenance would be worse -- since you would not only have to keep track of every ENTRY point for every source, but you would also have to ensure the linkage edit / binding process for every source included the appropriate aliases. The increased things to maintain and track indicate that maintenace for your specific situation would most likely be worsened by moving from static to dynamic calls, not to mention the initial overhead of adding the CANCEL statements to every program issuing a CALL.

dbzTHEdinosauer · Posted: Wed Sep 28, 2011 9:14 pm

my cent-and-a-half:

found little use for entry points in application code,
since moving to MVS from DOS last century.

as Bill inidcated,
rewrite the programs that have entry points,
and join the 21st century and do everything dynamically.

The only thing worse than a statically CALLed system
is some static and some dynamic.