Can't remember how I tested, but I know that I turn the async property off and I believe that I followed it with an XPath command. You are right, though, even with the XPath it's possible that the whole DOM had not been loaded. I know that the overall performance did not improve with clone, so the fact that the deferred DOM spread the load over time (if that's what happened) still worked for me.
-----Original Message----- From: Stanimir Stamenkov [mailto:[EMAIL PROTECTED] Sent: Wednesday, April 19, 2006 1:15 AM To: j-users@xerces.apache.org Subject: Re: best approach to whole document cloning in Xerces2? /Robert Houben/: > I ran extensive tests to see if clone would be faster (assumed it would, > at first). I found that reparsing the original file (assuming it hadn't > changed) was significantly faster than clone. If you have to serialize > first, you might lose that advantage but I seem to recall it was > significant. Could the faster reparsing be because of the deferred DOM implementation - have you tried measuring: parse an XML file and perform a no-op traversing through the DOM, then measure clone the document and perform a no-op traversing through the cloned DOM? -- Stanimir --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
