Re: When does manually allocated memory get released?

[Vern Hamberg <vhamberg@xxxxxxxxxxx>]

 Mark

I guess I'm of the school that says not to depend on hidden processes to 
clean up things. And I'm not sure about using some hidden thing 
sometimes, not in other situations.

In the handler I'm working on, I would not want to end the activation 
group, since I'm not the one who started it. I'm using *CALLER in the 
*SRVPGM that is the handler, as I recall.

I also don't know how many reads or other operations will be done 
between open and close. So I want to explicitly dealloc a particular 
based DS (that keeps the state between calls) in the event of an error 
and when closing the "file".

Just my way today!

Vern

On 10/8/2010 8:38 AM, Mark S. Waterbury wrote:
>    Unless you are talking about a very long-running server program, like
> a service that listens on a socket or a data queue, and is essentially a
> "never ending" job, there is really no point to ever doing a DEALLOC.
> This just adds unnecessary overhead.  The ILE heaps perform much faster
> if all you ever do is ALLOC, because internally it just advances the
> pointer into the free space, and does not have to worry about "holes".
>
> So, for most jobs, it is far more efficient to just wait until
> end-of-job and let the system take care of "throwing away" the entire
> heap, rather than incurring all of the overhead to issue DEALLOCate each
> chunk of storage you ALLOCated.