[XOM-interest] XOM performance compared

Wolfgang Hoschek whoschek at lbl.gov
Mon Jul 25 17:21:50 EDT 2005


Wrt. parsing, I just ran the ParseLoop.java performance measurement  
program included in dom4j-1.6.1 on my usual testbed (includes  
xerces-2.7.0). Overall, dom4j/SAX/xerces seems to be roughly 1.5  
times slower than xom-CVS on parsing/SAX/xerces, and dom4j/XPP pull  
parser is 5 times slower than dom4j/SAX/xerces.

As always, this is just a small little data point, and it may or may  
not apply to any given app use case, so please don't conclude too  
much from that, and please don't start flame wars.

Wolfgang.

On Jul 23, 2005, at 3:37 AM, Christof wrote:

> My co-worker did a very small test saying XOM 1.0 did take about  
> 10% more time to load/build/etc the document he tested it with. Of  
> course this is not a real performance test but thats what he went  
> with.
> I too doubt dom4j will be much faster but they are accustomed to it  
> and there should be not disadvantages by changing (advantages I see  
> in XOMs focus on correctnes. Simplicity for me is a big advantage  
> too but I myself did not use dom4j before).
>
> But I was wondering if XOM 1.1 will be a lot faster in some areas  
> after the discussions about performance optimizations on this list.  
> Being only a moderate I was not following everything in detail but  
> had the impression 1.1 should be much more optimized, is that true?  
> When is 1.1 expected as a final version?
>
> thanks
> Christof
>
>
> Wolfgang Hoschek wrote:
>
>> Would be interesting. I'm not aware of such a thorough study.  
>> Having  said that, I'd be somewhat surprised if they were more  
>> efficient. It  probably also depends on the main usage pattern.
>> Wolfgang.
>> On Jul 22, 2005, at 2:04 AM, csad7 at t-online.de wrote:
>>
>>> hello,
>>> are there any performance comparisons between lets say XOM and
>>> dom4j/JDOM etc?
>>>
>>> I try to persuade some colleges at work to use XOM instead of dom4j
>>> (which has bitten some projects with some nasty details mostly  
>>> issues
>>> about whitespace handling) but they question the performance of XOM.
>>>
>>> Googling did not bring any real help, so I'd think I'll ask here...
>>>
>>> thanks a lot
>>> christof
>>> _______________________________________________
>>> XOM-interest mailing list
>>> XOM-interest at lists.ibiblio.org
>>> http://lists.ibiblio.org/mailman/listinfo/xom-interest
>>>
>



More information about the XOM-interest mailing list