>  -----Original Message-----
>  From: Adam Turoff [mailto:ziggy@[...].com]
>  
>  On Mon, Jan 14, 2002 at 06:24:29PM -0500, Barrie Slaymaker wrote:
>  > At Matt's suggestion, there's a new machine that allows 
>  record oriented
>  > processing of XML, XML::SAX::ByRecord, along with a supporting SAX
>  > filter XML::Filter::DocSplitter.  X::S::ByRecord is 
>  documented below,
>  > feedback/testing/patches on any/all of this is quite welcome.
>  
>  I've been casually glancing at the list recently, and a couple of
>  mentions of XML::RAX started me thinking.
>  
>  Is there a standard idiom for converting a SAX machine into an
>  iterator?  That is, something like this:
>  
>  	my $xmlp = new XML::SAXIterator();
>  
>  	## set iterator options
>  	## connect to a driver that produces SAX events
>  	## use $xmlp to drive the entire process
>  
>  	while (defined ($result = $xmlp->next())) {
>  		## process the result object
>  		## could be a tree fragment, a la DOM or XML::Twig
>  		## could be some other type of data entirely
>  	}
>  
>  Writing event-based parsing code for lo these many years, my gut
>  feeling is that this isn't possible with XML::Parser or the SAX model
>  as it stands today.
>  
>  However, it would be with the addition of a pause() method and a
>  next() method.  The pause() method would be called within a handler
>  stack, and tell the parser to pause it's dispatch loop; at the end
>  of the current handler, it would exit from the dispatch loop,
>  returning control to the main program.  The next() method would
>  start parsing if it hasn't begun yet, and would continue parsing
>  from where the parser left off previously.
>  
>  A SAX parser would need to implement these methods, but SAX filters
>  would be building up object instance data during the parse, 
>  which would
>  be returned at the end of each iteration.  Thus, the filter would be
>  building up a twig, a Simple tree, a DOM-let, or some application 
>  defined structure (that's being unmarshalled from an XML stream).
>  
>  I *think* it's a simple hack to the SAX model.  And I *think* there
>  are some uses to iterator based processing, but I can't seem to 
>  come up with anything more than a variation of a record processor.

I think it would be good to discuss this on xml-dev (which I'm unsubscribed
from at the moment due to the noise level there), so that we can standardise
it for SAX for all languages. I'd certainly welcome a pause/resume system
(you wouldn't need next() if you had resume, I don't think).

Matt.

This e-mail has been scanned for all viruses by Star Internet. The service is powered by Mes
sageLabs. For more information on a proactive anti-virus service working around the clock, a
round the globe, visit: http://www.star.net.uk
_______________________________________________
Perl-XML mailing list
Perl-XML@[...].com
http://listserv.ActiveState.com/mailman/listinfo/perl-xml