Re: (binary?) DOM serialization

On Thu Jul 24, 2008 at 09:52:45PM +0200, Olivier Rossel wrote:
> 
> >  A
> > typical XML DOM library uses 5 times as much memory as the size of the XML
> > data read in.
> 
> Concerning that statement:
> For many years, "persistent DOM" implementations have proposed a
> clever binary file structure backuped by memory indexes that sticks to
> DOM API. This architecture helps infinitely when you deal with huge
> XML files.
> 
> The fact that these binary DOMs are not of common usage is probably
> only the lack of a (open and standard) reference implementation.

arent there all sorts of Acronymmy working groups on this. like OASIS or whoever. plus doesnt W3C have a binary-XML group since about a year now?

not to mention google's protocol buffers.. although i wonder if they store any DOM trees in those. proabbly not?

> 
> PS: please note the semantic difference between binary XML and binary
> DOM. A binary DOM requires as few parsing, mapping or translation as
> possible to be immediately usable by the library.

a (g)zipped (X)HTML serialization of DOM is pretty hard to beat, in terms of compatibility and 'future proofness', no?



firefox's 'Save as Webpage, complete' does a DOM serialize. but i havent investigated how much it really serializes back (eg, javascript on event bindings, etc)

which other browsers have a good DOM serialize. i cant really find it in opera. and im too lazy to compile webkit