Re: [xml] Using flyweights to reduce DOM size

[Daniel Veillard <veillard redhat com>]

On Sun, Mar 11, 2001 at 12:09:43PM +0000, Matt Sergeant wrote:
> I tried to parse a 54M XML file today (the gcide websters dictionary) to
> see what would happen. When it started to go into swap I had to kill it...
>
> MSXML uses flyweights [1] to reduce the size of thier DOM. They have some
> documentation on MSDN about how a MSXML DOM is generally about 0.9 to 1.1
> times the size of the original XML. I believe they also get a performance
> boost from this.
>
> Any thoughts on using the flyweight design pattern in libxml?
>
> [1] A well known design pattern, see
> http://www.mindspring.com/~mgrand/pattern_synopses.htm#Flyweight

  Yep, memory requirement is one of the limitation of libxml.
Libxml was designed initially as a document editing toolkit,
i.e. being able to save back everything and not optimized for
large data sets.
Changing those part would result in binary incompatible changes,
probably libxml3 when I manage to schedule such a beast.

Daniel

-- 
Daniel Veillard      | Red Hat Network http://redhat.com/products/network/
veillard redhat com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/