[jdom-interest] XPATH engine

[Kevin L Cobb]

Thanks for all the great and quick responces. Since I'm pretty wrapped
around JDOM at the moment, I think the next thing I'll try is Saxon's
Xpath implementation.  

-Kevin 

-----Original Message-----
From: Jason Hunter [mailto:jhunter at xquery.com] 
Sent: Wednesday, November 09, 2005 2:19 PM
To: Elliotte Harold
Cc: Kevin L Cobb; jdom-interest at jdom.org
Subject: Re: [jdom-interest] XPATH engine

Elliotte Harold wrote:

> Jason Hunter wrote:
> 
>> If performance matters, I'd use XQuery for it.  At MarkLogic we're 
>> able to get 100 XPaths per second against large data sets.  The speed

>> comes from the indexed nature of the storage (you can skip walking 
>> the tree).  It's not a JDOM plugin though, it's a separate server
process.
> 
> The speed gain there comes from the implementation, not the language. 
> There are XQuery implementations just as slow as JDOM's XPath, and 
> there are faster implementations of XPath as well. XQuery is not 
> fundamentally faster or slower than XPath/XSLT.

Yep, absolutely, and because XQuery is essentially a superset of XPath
that just makes sense.  :)

But in practical matters, the implementations that try hard to make
XPath fast tend strongly to be those that do XQuery, for various
environmental and competitive reasons.

If you want fast, you go where people are competing to be fast.

-jh-