Re: [xml] Dramatic performance gains with Libxml

[Liam R E Quin <liam holoweb net>]

On Sun, 2012-09-16 at 12:50 -0500, Ramon F Herrera wrote:
> Hello all:
>
> I am glad to report that after I replaced my previous XPath code with 
> libxml2, my application is running faster. In fact, the qualifier
> "Dramatic performance gains" is an *understatement*. This result is from 
> one of my many test cases:
>
> Processing with with Xerces + XQilla: 12 minutes
> Processing time with libxml:           3 seconds

Well, that's good news for you at least :-) but note that xqilla is not
optimized unless you use it with dbxml. Otherwise, it's one of the
slower XQuery implementations I've tried (and I've tried quite a few).

Sometimes one program is faster and sometimes another. Overall, libxml +
xsltproc is pretty fast for what it does, although I've measured a 100
times speedup moving to Saxon (XSLT 2 in Java) in some cases.

XQilla + dbxml gets its win because it only parses the XML once, and
then stores it in a database with an index (but you have to enable the
index or it won't be much faster). This is only a win if you process the
same XML document multiple times of course.

So you always have to look at the bigger picture

Liam

-- 
Liam Quin - XML Activity Lead, W3C, http://www.w3.org/People/Quin/
Pictures from old books: http://fromoldbooks.org/
Ankh: irc.sorcery.net irc.gnome.org freenode/#xml