Re: [xml] libxml2 XPath performances

[Daniel Veillard <veillard redhat com>]

On Sun, Nov 14, 2004 at 11:05:01PM +0000, Jose Commins wrote:
>       It is very illuminating to read the diverse comments on this thread 
> about XPath, as I have to keep in mind what potentially could be a vast 
> number of XML parsing which my program will perform in a production 
> environment, hence I have to keep performance in mind (Expat is out of 
> the equation for me, since libXML has the speed plus an excellent 
> feature set :o) .
>       What I imagined libXMl might have done on the initial 
>       validation/node creation list when reading an XML file is to perform a hash 
> tree of some sort, but it seems that is not the case.
>       With the Xpath logic aside, I reckon an LZW compression-like hashing 
> system could work quite well.  Any comments on implementing something 
> of this sort?

  Well, if you modify the tree, what happens to your hash ? How costly
is it to update it. libxml2 is primarily an editing toolkit, a lot of the
implementation choices have been driven by the ability to load/modify/save
easilly.

Daniel

-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/