After Parrot, what next?

[Brian Quinlan]

Mark 'Kamikaze' Hughes wrote:
>   Not even remotely.  SAX-like parsers make for really fast
> benchmarks,
> but they're totally useless for most real-world applications; even
> pull-based variants instead of the standard push-based ones
> suck.  What
> you *NEED* 255 times out of 256 is a DOM-like document tree.  And yet
> that forces your parser to have an optional cutoff point,
> which none of
> the standard implementations have.  My own fast, minimalist, DOM-like
> parser does that, but only after a bitter lesson from cruel reality.
>
>   Who lied to you and told you SAX was a good idea?

Are we talking about a DOM-like document tree being required for XML-based
languages? If so, XSLT is not an example where one is needed. The Xalan XSLT
implementation uses SAX to parse it's stylesheets. What other XML-based
language implementations have tried and failed to use SAX?