RE: Can files be opened local to a "subprocedure"?

I use it this way for some service programs.

I typically also use USROPN. I call it with *LOVAL in the first parm (regardless of its intended usage) to initiate the open and any setup I need.
Then it gets called with parms as needed during processing.
At the end, the first parm gets passed as *HIVAL to initiate a graceful close and cleanup processing if needed.

It's a methodology that has worked well for me.


-----Original Message-----
From: RPG400-L <rpg400-l-bounces@xxxxxxxxxxxxxxxxxx> On Behalf Of Vern Hamberg
Sent: Friday, October 8, 2021 9:03 PM
To: rpg400-l@xxxxxxxxxxxxxxxxxx
Subject: Re: Can files be opened local to a "subprocedure"?

Thanks, Roger, I'd forgot that - have never needed to use it, so,
brain-drain, perhaps!

Vern

On 10/8/2021 9:56 PM, Roger Harman wrote:

Unless you use the STATIC keyword....



-----Original Message-----
From: RPG400-L <rpg400-l-bounces@xxxxxxxxxxxxxxxxxx> On Behalf Of Vern Hamberg
Sent: Friday, October 8, 2021 6:00 PM
To: rpg400-l@xxxxxxxxxxxxxxxxxx
Subject: Re: Can files be opened local to a "subprocedure"?

Well, so far as I know, when a file is defined in a subprocedure,
basically you have a fresh open when the SP is called, and a close when
it exits - nothing is maintained between calls.

USROPN gets the same behavior - again, so far as I know.

Vern

On 10/8/2021 7:38 PM, James H. H. Lampert via RPG400-L wrote:

On 10/8/21 5:27 PM, Vern Hamberg wrote:

Now are maybe there issues of maintaining the position where you left
off in one of the files?

Of course there are. The record contents can be thrown out (the
recursion is very close to strict tail-recursion), but the position
and the EOF state can't. That is the whole point of the question, and
why opening it locally to the procedure invocation appears to be the
only way, short of caching the entire contents of the file.

And performance is not an issue. The file to be opened locally has an
absolute maximum of 55 records, and an absolute maximum of 49 records
that would satisfy the key conditions.

So far, in my tests, it appears that (at least at V6), simply moving
the F-spec to procedure level requires me to treat the file as
program-described, and define a structure to receive the record, but
treating externally-described files as program-described is old hat
around here. Apparently scatter-gather doesn't work for files declared
at procedure level?

--
JHHL