Re: Binding directory question

[Alan Campin <alan0307d@xxxxxxxxx>]

Problem is a lot deeper than that. A program has a single point of
entrance, a service program has multiple. There are ways to get around
this but all are about as ugly as they get. If service programs are
falling out of favor, how does one design with multiple entrance
points? If multiple small procedures are not being used, it points to
a design failure. Service programs are not being used correctly.

Also, program calls can be made using prototypes without fixed format
code. You can specify CONST on the parameter but their is nothing in
the receiver to enforce it.

d MyProgramCall...
d                            pr                      ExtPgm('MYPROG')
d   MyVariable...
d                                           7p 0    Const

 /Free

  MyProgramCall(3 * WorkVar);

 /End-Free


On Mon, Jan 26, 2009 at 9:04 AM,  <rob@xxxxxxxxx> wrote:
> In simple programming terms the problems that I have with external
> programs versus subprocedures in a service program are twofold.
>
> One, I can't do stuff like this in RPGLE
> RealVariable=MySubprocedure(3 * WorkVar);
> I have to something like this:
> TempVar=3 * WorkVar;
> call MySubprogram
> parm TempVar
> parm RealVariable
> Plus I have to convert the above into fixed format.
> Also, I have to be concerned with whether or not the subprogram mucked
> with TempVar since subprograms don't support CONST.
>
> Two, Since I cannot use the return variable capability of the above I
> cannot convert the subprogram into a UDF for use by SQL as in:
> http://faq.midrange.com/data/cache/185.html
>
> Rob Berendt
> --
> Group Dekko Services, LLC
> Dept 01.073
> Dock 108
> 6928N 400E
> Kendallville, IN 46755
> http://www.dekko.com
>
>
>
>
>
> From:
> <jvoris@xxxxxxxxxxx>
> To:
> "midrange-l" <midrange-l@xxxxxxxxxxxx>
> Date:
> 01/26/2009 10:26 AM
> Subject:
> Binding directory question
> Sent by:
> midrange-l-bounces@xxxxxxxxxxxx
>
>
>
> We have both types of programs in our shop, and for us, service programs
> are now falling out of favor.  Faster development and efforts in
> testability are our primary concern.
>
> I understand the push by IBM and others toward Binding directories and
> Service Programs. Yes, they are efficient.
>
> But for the programming behavior, where you are retrofitting and keeping
> the behavior consistent, it is extremely hard.  Managing binding is an
> added level of complexity that we do not see advantages to, and our
> users see little value in us addressing. (The ongoing "technical debt"
> of retrofitting and changing working programs is not something the user
> community wants to pay for now.)
>
> Because of good design and no copied code, where we have our functions
> are NOT defined in multiple places, but in one place only,  we really do
> not care if it is invoked as an EXTPGM or within Service Programs.  Not
> when response time is sufficient.
>
> Is there a response time problem with your applications?  Can you see a
> payoff in re-wiring them differently?  I believe you are taking on short
> term risk for little payback later.
>
> But the real question I think you should be asking is, Do the new and
> existing applications have automated tests to ensure code is written to
> match requirements?  And tests to prove the change did no harm, that we
> can prove anything we change is behaving the same way?   The failure to
> build these regression tests into system i code keeps us out of the
> Agile movement and from any true improvements in our codebase.  Until we
> change our programming habits in this way, it is all window dressing.
>
> In summary, I feel IBM and others push too hard into the area of binding
> and put too little effort into the area of testability.
>
> I am yet to be convinced that they are a "drastic" improvements from
> the plain old modular programming practices we institute, in the sense
> that as long as we have one program to handle one thing, and we have
> automated tests in place that prove the functionality again and again,
> we should be modify or refactor it without tests.
>
> So I too would be interested to hear about the experience and ROI that
> rewriting and/or re-wiring the application gives you.  I would love to
> be proved wrong.
>
>
> > from: David FOXWELL <David.FOXWELL@xxxxxxxxx>
> >
> > Jerry, I'd like to hear how you get on, it seems like we are just a step
> behind you.
> > >
> > Birgitta,
> > I'd be very worried about converting everything at once.
> > After having changed from 1 big bound pgm to one smaller pgm bound to
> several srvpgm, won't there be differences in behaviour?
> > What about things like Static variables, files staying open, etc?
> --
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing
> list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.
>
>
>
> --
> This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing list
> To post a message email: MIDRANGE-L@xxxxxxxxxxxx
> To subscribe, unsubscribe, or change list options,
> visit: http://lists.midrange.com/mailman/listinfo/midrange-l
> or email: MIDRANGE-L-request@xxxxxxxxxxxx
> Before posting, please take a moment to review the archives
> at http://archive.midrange.com/midrange-l.