Re: Basing Pointers (was X-Spec) -- MIDRANGE-L

Nelson,

I can't tell you what your problem is.  Last December, Midrange Computing 
published an article that describes a program that does essentially what you 
are describing.  It is not a trigger program, but I am using a version of this 
program as a trigger program, so I know it will work.  Let me know if you 
do not have access to this magazine and I will send you the relevant code.  
You should be able to use a prototyped call just fine.  Just reference a 
procedure pointer on the prototype for the EXTPROC like:

D pDynPrc           S                    PROCPTR
D Procedure       PR                 EXTPROC(pDynPrc)

David Morris

>>> "Nelson C. Smith" <ncsmith@gate.net> 05/12/99 08:31PM >>>

Jon, I've been having a problem with pointers that maybe you can shed some
light on.  It's more of a problem with a procedure pointer (I think),
although a couple of basing pointers are somewhat involved.

I have a common trigger program that is assigned to all files (so it never
has to be changed or moved) which calls a procedure in a service program,
which in turn checks a control file to find the name of another procedure
and service program which performs the actual trigger functions for the
specific file that fired the trigger.  It has been operating for some time
with no problems by using a hardcoded SELECT list to callp the procedure by
name.

I've been trying to implement a dynamic version using Barbara Morris's post
of about a year ago where she loaded a Procedure pointer to the procedure
name at run time using the API's to activate the service program where the
procedure lived and to retrieve the activation mark.  It works great except
for one small thing.  It can't seem to find its way back to the calling
procedure.  When I step all the way through the calling and called
procedures in debug, I get an error one step beyond the Return statement of
the called procedure saying that a pointer is not valid.

I can't really tell what pointer it is talking about, since the pointer that
got me to the procedure still contains the same address.  Is there some
pointer that the system sets up under the covers that points back to the
calling procedure or where I was in the calling procedure and is it possible
I'm trashing that pointer somehow?  How can I tell what pointer it is
talking about?  The joblog doesn't help any, and I don't get a dump.  The
debugger just displays the error and dies.

One additional item of information is that I am also passing two other
basing pointers as parms to the called procedure and they point back to the
original trigger buffer and another data structure of stuff I loaded.  I
don't think they could be trashing anything though because they work fine
now in the same procedures when called by the hardcoded Callp opcode. Those
two pointers are also still ok after I get the error.  The only other
difference in the dynamic version of the calling procedure is that there are
no prototypes in it since the procedures being called aren't known until
runtime.  I'm on V4R3.

Any ideas where I can look next?  I've hit the proverbial stone wall on this
one since I'm not really sure how the system gets back to a calling program.

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---