Mailing-List: contact cocoon-dev-help@xml.apache.org; run by ezmlm
Date: Mon, 17 Apr 2000 00:10:11 +0100
From: Paul Russell <Paul.Russell@uea.ac.uk>
To: cocoon-dev@xml.apache.org
Subject: Re: Cocoon 2.0: proposed battleplan
Message-ID: <20000417001011.B19662@nc1a.nca.uea.ac.uk>
References: <Pine.LNX.4.10.10004150122050.2459-100000@localhost.localdomain>
 <38FA45CD.190B3F17@apache.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.1.9i
In-Reply-To: <38FA45CD.190B3F17@apache.org>;
 from stefano@apache.org on Mon, Apr 17, 2000 at 12:59:25AM +0200
Sender: Paul Russell <paulr@nc1a.nca.uea.ac.uk>

On Mon, Apr 17, 2000 at 12:59:25AM +0200, Stefano Mazzocchi wrote:
> > But doing it before, the XML parser gets to do it internally. Doing it
> > after, we have to construct the DOM from SAX events. I reckon the former
> > would be faster... should I do some tests?
> 
> It seems to me Donald has a point here.

It's possible I'm missing the point here, but it's it true that any
parsing activity is fundamentally event based? If so, is SAX not
close to optimal?

I've written parsers for a number of projects in C (using yacc/bison)
and admittedly they've always ended up OO, but I suspect XML is possibly
the exception to this. We're talking about something that gets
arbitarily deep very very easily, and does not suffer from the complexity
problems that plague other language grammers.

I don't know, I could be completely wrong, but I don't see us obtaining
more than an extra one to two percent performance increase in not using
SAX. Personally, I'd rather have the flexibility of working with event
streams, if the application allows.


Paul