Re: ILE bug or "feature"?

[Douglas Handy <dhandy1@xxxxxxxxxxxxx>]

Scott,

>This implies that when the module is loaded into memory that it's not
>initializing its variables to NULL, as I believe that it should.

These were *Entry parameters, so it can't initialize the variables.  When you
use a *Entry PLIST, it expects to find pointers for each PARM on the call stack.
Since a previous invocation of TEST1 had put pointers there and the call to
TEST2 had not replaced any of them, the pointers were still sitting on the
stack.

>It seems to me that the addresses of the parameters should be *NULL if
>they weren't ever SET by passing parms to the procedure.  

XTEST did not use a prototyped call to TESTn, so it has no way of knowing that
TEST2 may be expecting 5 parameters.  All RPGLE programs will show a parameter
range of 0 minimum to 255 maximum in the object description.  (Do a DSPPGM and
roll down 3 times to see this.)  So do you think it should have called TEST1
with 5 pointers followed by 250 null pointers, and TEST2 with 255 null pointers?

At 16 bytes per pointer, that would seem to be a waste of about 4K per call
level just to hold null pointers.  How significant is that in today's
architecture?  I don't know.

>The fact that
>they happened to point to the same area of memory that the previous
>procedure pointed to seems like an extraordinary coincidence!

Not to me.  That's just the way that stacks work.  I'd be more surprised if it
*didn't* behave the way Mark experienced.

This just exemplifies the advantages of using prototyped calls (when possible)
and using %parms to check how many parameters were actually passed.

Doug
+---
| 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
+---