xsl-list
[Top] [All Lists]

RE: Performance optimization and avoiding Out of Memory errors

2002-09-06 03:09:43


-----Original Message-----
From: owner-xsl-list(_at_)lists(_dot_)mulberrytech(_dot_)com
[mailto:owner-xsl-list(_at_)lists(_dot_)mulberrytech(_dot_)com]On Behalf Of 
Lewden Romain

Hello Lewden,

I'm working with quite important XML Data Base ( 10 Mo) and trying to
extract information and transform it into HTML files and rtf format.
I work with Xt processor (release 19990511)
I have two big problems :
1)time of genration is very important
I create a predocument xml of 10 Mo (with 10 chained requests about 6,000
objects)
This step takes 3 hours, with a station of 512 RAM Memory

first off, with large processing scenarios, as you have....have you
considered using a SAX approach ?

what is the format of this pre-document.....I assume it is XML. what are you
doing in this pre-step ?

have you thought of using an XML native repository like xindice or
Tamino....seems like you also benefit from db style optimisations dealing
with so much data.


Then I create in HTML a file for each object; it takes me 14 hours.
it is a recursive call :
<dog level="0" name="Toto1">
      <dog level="1" name="Toto1">
              <dog level="2" name="Toto1"
                      <dog level="3" name="Toto1">
                              <cat>ieodfli</cat>
                              <cat>ieodfli</cat>
                      </dog>
              </dog>
              <dog level="2" name="Mydog">
                      <dog level="3" name="Toto1">
                              <cat>ieodfli</cat>
                              <cat>ieodfli</cat>
                      </dog>
              </dog>
      </dog>
</dog>
I've a template for "dog", with recursive calls, and if I'm in level "2" I
create a separate HTML page. I've 6000 elements at level 2 and it takes me
10 hours. How can I optimize this time ?

hmmmm, why do you store level ? I think that if u described your entire
process that we could come up with something optimised.....otherwise the
structure of your XML ( is this lorem ipsum, or are u really doing something
with dogs and cats ?) is slightly confusing.

without seeing your template, or use of keys, etc.....it is hard to comment.


2) When I work in rtf, I can have out of memory errors, because I
write less
files but they become huge (9 Mo octets).
All the output file is kept in memory in memory until the stylesheet is
performed.
How can I introduce an option to force my program to write "in
real time" in
the output file ?

my gut feeling, is that you should investigate Saxon (
saxon.sourceforge.net ) as it has some switches to assist in large file
processing.....and it is also a java XSLT processor, which will integrate
with your environment easily.


gl, jim fuller


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list



<Prev in Thread] Current Thread [Next in Thread>