[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: Parsing Performance

• To: <xml-dev@l...>
• Subject: RE: Parsing Performance
• From: David Megginson <david@m...>
• Date: Mon, 24 Feb 2003 08:27:42 -0500
• In-reply-to: <000901c2dbfc$a11821b0$6c64a8c0@DOGGIE>
• References: <15961.7999.784906.56668@m...><000901c2dbfc$a11821b0$6c64a8c0@DOGGIE>

Martin Soukup writes:

 > > Is it some kind of indexing engine, or are you reading live datafeeds?
 > 
 > Live data feeds. Yes, I think you might be right, but what option
 > do I have other than SAX?

First you should determine how big a hit you're actually taking from
the exceptions in the context of your full application.  This is far
from perfect, but it might give you a rough idea:

1. Take one of your typical-length XML documents.

2. Create a trimmed-down version of the same document, with half of
   the content missing.

3. In your current system, time how long it takes to process the
   first document (say) 10,000 times, terminating halfway through each
   time by throwing an exception from a SAX handler (at the point
   where the trimmed-down document ends).

4. Next, time how long it takes to process the second document 10,000
   times without premature termination.

This test is slightly misleading, because step #4 might include a
little extra overhead at the end of the XML document (checking for
well-formedness), but it should give you a fairly reasonable idea.  My
guess is that the exception overhead easily gets lost in the noise --
XML parsers do a *lot* of work for every event that they report.  1000
times practically nothing is still practically nothing.


All the best,


David

-- 
David Megginson, david@m..., http://www.megginson.com/

• References:
  • RE: Parsing Performance
    • From: David Megginson <david@m...>
  • RE: Parsing Performance
    • From: "Martin Soukup" <martin@d...>

• Prev by Date: RE: XML for RFCs
• Next by Date: RE: XML for RFCs
• Previous by thread: RE: Parsing Performance
• Next by thread: Re: Parsing Performance
• Index(es):
  • Date
  • Thread