Re: KHTML in GNUStep?

[Chris B. Vetter]

On Fri, 17 Jan 2003 11:52:18 +0100
Helge Hess <helge.hess@skyrix.com> wrote:
> Chris B. Vetter wrote:
> > Well, I've been thinking about writing a DOM parser in ObjC for a
> > while; that would include XML, HTML, ECMA and CSS, and eventually a
> > rendering'engine' as well. Nothing tangible yet, though...
> We just released our "DOM implementation" in the contrib directory of 
> ftp.gnustep.org. I'm not entirely happy with the current
> implementation, but it might be a good starting point. So instead of
> writing a DOM from scratch you could help improving that piece (eg the
> performance and compliance of the DOM tree implementation).
> At least it can already build a DOM structure from various sources, eg
> HTML files, XML files, iCal files, plist files, ...
> We also have a bridge to the SpiderMonkey JavaScript engine (your ECMA
> point) that can also access the existing DOM. Maybe we also release
> this package later.

It certainly looks interesting and might indeed be a starting point -
however, I'm personally not particularly fond of SAX, for several
reasons. One, it was primarily designed for writing XML parsers only,
which isn't a big problem if that's all you want/need - BUT (and that's
reason two) - SAX1 is NOT entirely DOM compliant, and SAX2 differs
significantly from SAX1, even if it will be somewhat backward
compatible. So, as far as I know, only if/when the DOM3 specs will be
out there there's chance SAX and DOM will "merge". And third, SAX
takes documents as "streams" instead of a tree, as defined in DOM, which
means it's much harder to do forward/backward references to objects.

But that's just my personal view.

-- 
Chris