Hello again Henry,
To be able to responde to your questions I made a new test, and this time I
captured the sun prstat from my machine to give you an idea of my issues.
The test I run was without INCREMENT option and using the file I talked you
about previously.
I run the command: java -d64 -cp .:xalan.jar:xerces.jar
org.apache.xalan.xstl.Process -in ubeefile.xml -xsl GPRS2FTX.xsl -out out.file
Follows a copy of prstat command output on a sun sparc machine:
PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP
23906 geneva 918M 753M cpu3 10 0 0:00.42 12% java/23
23906 geneva 918M 753M cpu3 10 0 0:00.42 12% java/23
23906 geneva 1109M 1047M cpu2 25 0 0:01.59 9.4% java/23
23906 geneva 1109M 1047M cpu17 10 0 0:01.06 9.5% java/23
23906 geneva 1109M 1048M cpu18 50 0 0:02.42 9.5% java/23
becomes stable at this point
...
after some minutes the process aborts with:
Exception in thread "main" java.lang.OutOfMemoryError: Java
heap space
Follows also some important information:
$ wc -l ubeefile.xml
203330
Hope this helps clear what I am experiencing
Best regards
Higino Silva
-----Original Message-----
From: Henry Zongaro [mailto:[EMAIL PROTECTED]
Sent: quarta-feira, 4 de Maio de 2005 13:56
To: Higino Silva; Higino Silva, Eng�
Cc: [EMAIL PROTECTED]
Subject: RE: Help using the Xalan mailing lists
Hi, Higino.
Higino Silva, Eng� <[EMAIL PROTECTED]> wrote on 2005-05-03 01:37:59 PM:
> Thanks for your feedback. One question, can you see in your tests the
> memory increasing over time like me? It becomes more "readble"
> if you turn INCREMENTAL on.
I'm not sure what you mean by "increasing over time." Do you mean that
the amount of memory consumed increases from one transformation to the next, or
do you mean that the amount of memory consumed increases gradually during a
single transformation?
If you mean during a single transformation, that would be expected.
The INCREMENTAL feature builds a DTM representation of the input document
gradually while the transformation is running, so it's possible for a
transformation to start producing output while it's still building a
representation of the input document. If you mean that the amount of memory
consumed increases from one transformation to the next, and that garbage
collection doesn't seem to return all the memory, then that might represent a
memory leak bug, and we would need to investigate further.
Thanks,
Henry
------------------------------------------------------------------
Henry Zongaro Xalan development
IBM SWS Toronto Lab T/L 969-6044; Phone +1 905 413-6044
mailto:[EMAIL PROTECTED]
