[From nobody Fri Aug  6 17:07:27 2004
Return-Path: <kgillies@gemini.edu>
Received: from alias.acm.org ([199.222.69.90])
	by bissell.mail.mindspring.net (Earthlink Mail Service) with ESMTP id
	16Kzgk1FD3Nl3rE0
	for <servlets@mindspring.com>; Mon, 11 Mar 2002 18:31:28 -0500 (EST)
Received: from cygnus.hi.gemini.edu (128.171.188.208)
	by alias.acm.org with MERCUR Mailserver (v4.01.11 OTAtMjE3Ni00NjIw)
	for <servlets@mindspring.com>; Mon, 11 Mar 2002 18:27:54 -0500
Received: from localhost (dhcp1-132 [192.168.1.132])
	by cygnus.hi.gemini.edu (Pro-8.9.3/Pro-8.9.3) with ESMTP id NAA01868
	for <jhunter@acm.org>; Mon, 11 Mar 2002 13:28:15 -1000 (HST)
Date: Mon, 11 Mar 2002 16:28:14 -0700
Mime-Version: 1.0 (Apple Message framework v481)
Content-Type: text/plain; charset=US-ASCII; format=flowed
Subject: JDOM ?
From: "K.Gillies" <kgillies@gemini.edu>
To: jhunter@acm.org
Content-Transfer-Encoding: 7bit
Message-Id: <A8965ECC-3547-11D6-A411-00039340AA82@gemini.edu>
X-Mailer: Apple Mail (2.481)
X-Mozilla-Status2: 00000000

Hi Jason,
   I apologize for writing you directly.  I'm not a regular watcher of 
the JDOM
lists although I'm use JDOM -- and I like it!  So if you like, please 
give me
advice.

   I have an application that reads many XML files that must be validated 
to
several DTDs.  Someone suggested that creating XML parsers and 
configuring
every time a new file must be parsed was rather expensive and that 
parsers
should be cached like JDBC connections are often cached.  A new parser
isn't needed for every document.  I didn't do any measurements since
it seemed like a reasonable statement.

   So I built a class to provide a central cache of SAXBuilders (one to
start with) and allowed clients of the cache to add EntityResolver 
instances
to locally return entities.  My single EntityResolver just goes through
all the client EntityResolvers one by one.

   All was good until I tried parsing multiple documents, then I got a
JDOMException "Error in building: Stream closed"

   I then went to the SAXBuilder source to look around.  What I found is 
that
SAXBuilder always creates a new parser in the build method.  So all my 
efforts
were misplaced.  It appears that SAXBuilder can't be resued.

   Do you have any comments on this?  Shouldn't I be concerned about the 
cost of
creating/configuring parsers every time a new document must be parsed?  
One
case where this seems important is a web service that needs to handle XML
messages (like a service using SOAP).

   Is there a way to reuse a SAXBuilder and not create a new parser for 
every
document?

   Thanks for any advice you have.

Kim Gillies


]