Ptr to Fast Well-Formed XML parser (Java)?

Michael.Kay at Michael.Kay at
Tue Jan 19 18:43:11 GMT 1999

> Last year in May, Mike Kay made a parser benchmark with a 
> simple SAX 1.0 application. The bench was between AElfred, Lark, MSXML, 
> xml4j and XP and at the time XP was the fastest. Perhaps he could do the 
> benchmark again?

It so happens I did some recent measurements (different data file) as

SAX parsers:
xp         4387 ms
oracle     7862 ms
xml4j      7771 ms
sun        3736 ms

DOM implementations:
sun        9634 ms
xml4j     11677 ms
oracle     9784 ms
docuverse 10685 ms (using xp parser)

So SUN is looking pretty good at the SAX level, but at present its licensing
terms are rather restrictive. For DOM products, they're all remarkably

I think that the oracle and ibm SAX parsers are slower because they build a
parse tree even if you don't need it. Can anyone confirm this?

I was intending to complete the survey by running other parsers such as
aelfred, lark, datachannel, silfide; but either I've had problems getting
them to run, or I just haven't found the time.

Unfortunately, once again, I'm using a confidential data set, in this case
one from a client.
Sorry! Its size is 690Kb, and its internal structure is quite rich.

The figures are obtained directly from the com.icl.saxon.Renderer main
program within SAXON, selecting different parsers by editing the file. So you can easily run the same tests on your
own data files. This program includes the basic SAXON processing (which
builds a stack and selects an element handler for each element based on its
type), but it uses a null element handler for all elements.

Mike Kay

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at
Archived as:
To (un)subscribe, mailto:majordomo at the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at

More information about the Xml-dev mailing list