Hi Mr Iff

SAX has pluses and minuses and these will obviously depend
on your perspective.  Comparing SAX to XML::Simple is a bit
of an apples and oranges comparison though.  

If you're looking for something more powerful than XML::Simple,
then a DOM module like XML::XPath, XML::LibXML or (at a pinch
XML::DOM) is a better comparsion.

If you're looking for something to compare SAX to, then think
of the XML::Parser native API where you define event handlers.

SAX is a huge advance over the XML::Parser Handler API for a
number of reasons (including):

 - pluggable - if your code is written to the SAX API you
   can use any SAX parser without changing your code

 - modular - you can write simple (single function) filter
   classes and string them together for the desired result

 - well supported - there are lots of existing SAX filters,
   handlers, drivers and glue modules like XML::SAX::Base
   and XML::SAX::Machines

 - flexible - your data source does not even need to be an
   XML document (eg: you can drive your SAX pipeline from
   a database query

 - namespace support (SAX2 only) which is consistent across
   different parsers

However, as you have discovered SAX is not a panacea (does 
anyone ever use the word 'panacea' without the 'is not a'
prefix?).  If you're coming from an XML::Simple viewpoint,
then the first thing you notice about SAX is that you have
to write quite a lot of code to use it.  It doesn't need to
be complex code and the API is not extensive, but you do
have to write rather a lot of it.

I'm not entirely clear on what you're trying to do with
namespaces.  Do you want your hashref keys to be in Clarkian
notation eg: '{http://purl.org/dc/elements/1.1/}date' or
do you want to normalise the prefixes used eg: 'dc:date'?

XML::Simple version 1.08_01 will do the Clarkian notation
for you if you set nsexpand =>  1 (as long as you're using
a SAX parser).

You could achieve the normalisation effect with a SAX filter
(does anyone have an XML::SAX::NSNormalise they'd like to
contribute to CPAN?).  And if you use XML::Simple as the
handler on the end of the SAX pipeline you get the best of
both worlds.

Hope that helps
Grant

===============================================================
Grant McLean        BearingPoint Inc - formerly The Web Limited
+64 4 495 8250           Level 6, 20 Customhouse Quay, Box 1195
gmclean@[...].biz                Wellington, New Zealand




>  -----Original Message-----
>  From: Morbus Iff [mailto:morbus@[...].com]
>  Sent: Tuesday, October 08, 2002 9:02 AM
>  To: Perl-XML@[...].com
>  Subject: What's So Great about SAX? (ie. Future Indecisions)
>  
>  
>  
>  I have a popular (nearly 40,000 downloads) piece of software. It's 
>  available on Mac (Classic and OS X), Windows (all), and most 
>  Linux systems 
>  I've tested it on. Currently, it's using XML::Simple to build various 
>  formats of XML into a hash/tree data structure. It's been 
>  doing that for 
>  the past two years now.
>  
>  I've gotten some other opinions on this, but
>  roughly, I want to hear from the rest of the group:
>  
>    - is there any benefit from moving from
>      XML::Simple to something like XML::SAX?
>  
>  Keep in mind:
>  
>    - expat is the only xml parser currently fully crossplatform
>      for all my needs. it's also easily included into an
>      .exe or application file (for Mac or Windows).
>  
>    - I'm using XML::Simple, and I only want things in a hashref.
>  
>    - I've been hacking up namespace recognition code to go
>      with the hashref XML::Simple uses. This was the "last
>      straw" that made me look into XML::SAX again.
>  
>    - I rarely use XML::Parser's API for anything crucial. The nearest
>      is to stop XML::Simple from creating a hashref of the entire
>      alphabet - I dump into XML::Parser to create a tree of only
>      letter 'b', for example.
>  
>    - I'm a fan of readability. This attempt at SAX parsing for
>      the format I'm interested in (RSS) is "ooky" and not
>      code shui, in my head:
>  
>       http://aaronland.info/weblog/archive/4637
>  
>      A hearty thanks to Aaron for exploring that for me.
>  
>    - Common API portability isn't crucial to me, since
>      XML::Simple figures all that crap out for me (thanks, Grant!).
>  
>  Thoughts?
>  
>  
>  -- 
>  Morbus Iff 
_______________________________________________
Perl-XML mailing list
Perl-XML@[...].com
To unsubscribe: http://listserv.ActiveState.com/mailman/mysubs