Memory

[David Mechner <fo...@mechner.com>]

Hi,

I'm using FOP (along with JFreeChart, Batik, and Xerces) to generate
pdf reports, and it's working great - many thanks to the authors!

However, I'm running into some stability problems. I'm trying to
generate about 1000 4 page documents, and there seems to be a slow
leak in memory - giving the jvm a maximum heap of 1GB, the jvm runs
out of heap after about 350 documents.

I coded the report generation to run in a single process so I could
reuse common parts of the report, and use Java DOM trees as
intermediate representations in the Db-XML-Xerces-FO-Fop chain on the
theory that avoiding IO would be most efficient.

Does anyone have any insight into where the leak might be, or whether
it's just a waste of effort to try to keep this in a single process
rather than writing a batch program to invoke multiple times so I
won't run out of memory (that just feels like defeat, but I'm willing
to compromise).

Thanks,

-David

Re: Memory

["J.Pietschmann" <j3...@yahoo.de>]

David Mechner wrote:
> However, I'm running into some stability problems. I'm trying to
> generate about 1000 4 page documents, and there seems to be a slow
> leak in memory - giving the jvm a maximum heap of 1GB, the jvm runs
> out of heap after about 350 documents.
> 
> I coded the report generation to run in a single process so I could
> reuse common parts of the report, and use Java DOM trees as
> intermediate representations in the Db-XML-Xerces-FO-Fop chain on the
> theory that avoiding IO would be most efficient.
> 
> Does anyone have any insight into where the leak might be,

FOP caches all external-graphics. If you are using
SVG for graphics, you can try to embed them using
fo:inline-foreign-object.

J.Pietschmann

Re[2]: Memory

[David Mechner <fo...@mechner.com>]

Xalan-j_2_4_D1. I'm also using  batik-1.1.1 and fop-0.20.3.

-David


You wrote:
JEK> What jaxp implementation are you using?

JEK> Jason





JEK> -----Original Message-----
JEK> From: David Mechner [mailto:fop@mechner.com] 
JEK> Sent: Wednesday, August 28, 2002 4:25 PM
JEK> To: fop-user@xml.apache.org
JEK> Subject: Memory

JEK> Hi,

JEK> I'm using FOP (along with JFreeChart, Batik, and Xerces) to generate
JEK> pdf reports, and it's working great - many thanks to the authors!

JEK> However, I'm running into some stability problems. I'm trying to
JEK> generate about 1000 4 page documents, and there seems to be a slow
JEK> leak in memory - giving the jvm a maximum heap of 1GB, the jvm runs
JEK> out of heap after about 350 documents.

JEK> I coded the report generation to run in a single process so I could
JEK> reuse common parts of the report, and use Java DOM trees as
JEK> intermediate representations in the Db-XML-Xerces-FO-Fop chain on the
JEK> theory that avoiding IO would be most efficient.

JEK> Does anyone have any insight into where the leak might be, or whether
JEK> it's just a waste of effort to try to keep this in a single process
JEK> rather than writing a batch program to invoke multiple times so I
JEK> won't run out of memory (that just feels like defeat, but I'm willing
JEK> to compromise).

JEK> Thanks,

JEK> -David

RE: Memory

["Jason E. Kafer" <je...@ritvax.isc.rit.edu>]

What jaxp implementation are you using?

Jason





-----Original Message-----
From: David Mechner [mailto:fop@mechner.com] 
Sent: Wednesday, August 28, 2002 4:25 PM
To: fop-user@xml.apache.org
Subject: Memory

Hi,

I'm using FOP (along with JFreeChart, Batik, and Xerces) to generate
pdf reports, and it's working great - many thanks to the authors!

However, I'm running into some stability problems. I'm trying to
generate about 1000 4 page documents, and there seems to be a slow
leak in memory - giving the jvm a maximum heap of 1GB, the jvm runs
out of heap after about 350 documents.

I coded the report generation to run in a single process so I could
reuse common parts of the report, and use Java DOM trees as
intermediate representations in the Db-XML-Xerces-FO-Fop chain on the
theory that avoiding IO would be most efficient.

Does anyone have any insight into where the leak might be, or whether
it's just a waste of effort to try to keep this in a single process
rather than writing a batch program to invoke multiple times so I
won't run out of memory (that just feels like defeat, but I'm willing
to compromise).

Thanks,

-David