Mailing-List: contact user-help@poi.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "POI Users List" <user@poi.apache.org>
Received-SPF: pass (athena.apache.org: domain of lists@nabble.com designates
 216.139.236.158 as permitted sender)
Message-ID: <20794395.post@talk.nabble.com>
Date: Tue, 2 Dec 2008 07:57:01 -0800 (PST)
From: Rob Y <ryampolsky@yahoo.com>
To: user@poi.apache.org
Subject: Re: Using POI via JNI
In-Reply-To: <a644352c0812011826t78870c82n5a1d9a01ff35f87c@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
References: <20774758.post@talk.nabble.com> <20775023.post@talk.nabble.com>
 <a644352c0812011826t78870c82n5a1d9a01ff35f87c@mail.gmail.com>


Josh Micich wrote:
> 
> Hello Rob,
> 
> So you might be stuck with leaving the JVM permanently loaded.  This
> may not be such a bad thing if you application is multi-threaded and
> has the potential for concurrent calls to POI.  Multi-threading would
> be harder to support assuming short-lived JVMs.
> 
> cheers,
> Josh
> 

Leaving the JVM loaded is fine so long as my 'print to excel' library is the
only thing that uses Java in the app.  But if the calling app wanted to use
Java itself, there'd be a conflict.  At very least, it'd want to start up a
JVM with a different classpath than I use, since I've got my code and poi in
.jar files that make up my classpath.

My app's structured Apache-like as a transaction processing system with a
pool of shared processes as opposed to a single multi-threaded process.  The
second time a user goes to create an XLS, they may be running in a different
process than the first time, so having the JVM preloaded is less of a
benefit and more of an overhead.  I think it'd be better to just exit my
process after creating the XLS and let the system dispatcher launch a new
copy as needed.  Luckily, I have that option, but it's pretty incredible
that Sun's had this DestroyJavaVM thing in their documentation all along and
still haven't implemented it properly.

Does anybody know how Apache Tomcat deals with this?  Presumably it's got a
pool of processes that need to run different Java code on different
invocations.  I guess if they all use the same classpath, then the same JVM
could service them all.  Is that the trick?

Rob

-- 
View this message in context: http://www.nabble.com/Using-POI-via-JNI-tp20774758p20794395.html
Sent from the POI - User mailing list archive at Nabble.com.


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@poi.apache.org
For additional commands, e-mail: user-help@poi.apache.org