RE: Excel/POI - Please Help The Drowning Man

The end is in sight! (Though the end, in this case, may involve me hanging 
from the nearest rafter.)

I changed to Scott's new version of the hssf service program. And, as John 
Joiner pointed out, the "wwFile" definition in "hssf_open" was incorrectly 
defined as "like(jFileOutputStream)" which I changed to FileInputStream.

Now there's one more thing wrong but at least it's CONSISTENT. 

If I run my 'read the s/sheet' job interactively, it WORKS EVERY TIME. 
If I run it in batch, IT FAILS EVERY TIME on the 'open' (Java 
IOException)! 
I've tried CPYENVVAR(*YES) and/or ALWMLTTHD(*YES) on the SBMJOB, all to no 
avail. 

Does anyone know what, in a batch job's environment, could be so different 
from the interactive that the batch job would fail every time???

You guys have been great. Thanks so much.

Arthur J. Marino
Southern Container Corporation
(631) 231-0400 x133



"John Joiner" <john@xxxxxxxxxxxxxx> 
Sent by: rpg400-l-bounces@xxxxxxxxxxxx
08/28/2006 02:06 PM
Please respond to
RPG programming on the AS400 / iSeries <rpg400-l@xxxxxxxxxxxx>


To
"'RPG programming on the AS400 / iSeries'" <rpg400-l@xxxxxxxxxxxx>
cc

Subject
RE: Excel/POI - Please Help The Drowning Man

I've been trying to reproduce this problem so that I can fix my service 
program for everyone who uses it.  But, so far, I've been unable to 
reproduce the problem you describe.
One problem I've had with the hssf_open() procedure is that it declares
'wwfile' as type 'jFileOutputStream' and passes this variable to 

procedure

'new_POIFSFileSystem' to open an existing workbook. 

Scott,

When I originally ran your UPDDEMO program, it kept bombing while trying 
to
open the workbook with a java IOException in the POIFSFileSystem
constructor, as the constructor expects a FileInputStream, but a
FIleOutputStream is being passed instead.

I changed the declaration of 'wwfile' to jFileInputStream' and got around
this problem.

This did not cause the scenario described in the original post, but the 
next
problem I ran into did.

Once I was able to open the spreadsheet, the program bombed trying to 
create
the 'cell' object from the 'row' object in the following line:
cell = HSSF_GetCell(row: 1);

In my case, it bombed because the previous statement, which creates the
'row' object, returned a null reference for 'row'. Passing a null 
reference
to the HSSF_GetCell procedure caused all finds of problems, including the
problem described in the original post. 

Looking at the joblog, I saw the following errors for SRVPGM QJVAJNI after
the call to HSSF_GetCell:
MCH74A5 - Java Virtual Machine has ended with a reason code of 2.
MCH3601 - Pointer not set for location referenced

Each subsequent call to 'hssf_begin_object_group' results in the following
errors:
JVAB55C - Unable to create Java Virtual Machine because of reason 5 (JVM
already started in process)

As you mentioned, I can sign off and then back "resets" everything.

Hope this helps.

JJ

-----Original Message-----
From: rpg400-l-bounces+john=thejoiners.net@xxxxxxxxxxxx
[mailto:rpg400-l-bounces+john=thejoiners.net@xxxxxxxxxxxx] On Behalf Of
Scott Klement
Sent: Monday, August 28, 2006 11:03 AM
To: RPG programming on the AS400 / iSeries
Subject: Re: Excel/POI - Please Help The Drowning Man