Re: FOUND IT, Re: WEIRD PROBLEMS, Re: Front-ending Java for RPG -- Anybody here familiar with it besides Scott?

On 2/13/2013 4:26 PM, James H. H. Lampert wrote:

stmt = rs_getStatement(rs);
if (stmt <> *NULL);
stmt_close(stmt);
endif;
DeleteLocalRef(JNIENV_P: rs);
if (stmt <> *NULL);
DeleteLocalRef(JNIENV_P: stmt);
endif;
rs = *NULL;

Would anybody happen to know whether or not it's OK to change the order
of the DeleteLocalRef calls (i.e., do "stmt" before "rs")? This would
allow me to avoid having two identical IF statements, but I fear it
might also create a hidden flaw that might not show up in tests. If I
can help it, I'd rather not send out a product with a memory leak (or
worse)!

It feels to me like it would make more sense to delete the reference to stmt before deleting the ref to rs. But for sure I think you should set rs to *null right after you delete the reference to it, to avoid any potential of using a stale object reference.

I don't think the order matters one way or the other. DeleteLocalRef doesn't tell Java to free the object. It just tells Java that you don't need _your_ reference to the object any more. Java can only free an object when there are no more references to it anywhere, whether in native code or within the Java objects themselves.

By the way, maybe you just didn't show it, but you should be setting stmt to *null too.

I don't see any reason you couldn't code it like this, keeping all the stmt code together.

stmt = rs_getStatement(rs);
if (stmt <> *NULL);
stmt_close(stmt);
DeleteLocalRef(JNIENV_P: stmt);
stmt = *null;
endif;
DeleteLocalRef(JNIENV_P: rs);
rs = *NULL;