[
Lists Home |
Date Index |
Thread Index
]
Elliotte Harold wrote:
> Dennis Sosnoski wrote:
>
>
>> Not sure what you're basing this on, Rich, but it doesn't match what
>> I've seen. See the results at
>> http://xbis.sourceforge.net/performance.html, for instance, where
>> generating text XML from a parse event stream takes on average about
>> the same amount of time as parsing the text to get back the parse
>> event stream
>
>
> That sounds suspicious. Parsing has a lot more overhead than just
> writing out text strings. I suspect the real comparison here may be
> between a system that does full well-formedness checking and one that
> doesn't. Many systems can blast XML out pretty quickly using printf or
> equivalent. They don't necessarily need to call a method for each tag
> or text node or check well-formedness on output.
>
Perhaps so, though the types of data you can use with the printf
approach are severely limited. Generally text from other types of
documents, or from a database, can't be handled this way because it
includes characters that need to be escaped. Besides, I don't know how
much you actually gain with this approach; at least in Java, I suspect
it wouldn't be all that much.
I think the data binding comparisons are very relevant for a look at
real-world type usage. Some frameworks are exceptions (such as Castor,
which is much slower on input than JAXB but about the same speed on
output), but in general output speed is only slightly faster than input
speed.
- Dennis
--
Dennis M. Sosnoski
Enterprise Java, XML, and Web Services
Training and Consulting
http://www.sosnoski.com
Redmond, WA 425.885.7197
|