Re: JDOM performance

["David Morris" <David.Morris@xxxxxxxxxxxxx>]

Keith,

JDOM is a great tool that simplifies working with XML documents.
As the name suggests, it builds a DOM tree that includes the entire
file in memory. Internally JDOM stores your document in a linked
lists. JDOM is pretty efficient in how it stores your data however,
at this point there is some redundancy. JDOM is being converted
to use the flyweight design pattern but I don't know how far along
that is. I have profiled JDOM and it is pretty easy on memory --
much less memory intensive than Xalan. SAX will use much less
memory if you process your document top to bottom and can
work from events. SAX is pretty easy to use but you cannot
manipulate documents as easily.

David Morris


>>> keith_mccully@wunderman.co.uk 11/26/02 06:29AM >>>
This is a multipart message in MIME format.
--
[ Picked text/plain from multipart/alternative ]
I'm evaluating the Beta 8 version of JDOM on the iSeries. So far I've
written a few test classes and am very impressed with its capabilities
for
reading and writing XML documents. However, I've read that JDOM loads
the
whole document tree into memory and keeps it there until output. I'm
not
sure how the iSeries memory management handles this but there seems
reasonable cause for concern.

I am testing transformations between XML and DB2/400 in both
directions.
So far, using small test files which have worked fine but later I will
need to handle files containing several thousand records.

Has anyone got any experience using JDOM with large files? I am
interested
in the performance and if JDOM is slow (and slows down the machine)
then I
will consider going back to SAX and DOM but coding with JDOM is a
whole
lot easier!

Thanks,

Keith