RE: Sticks, carrots, real XML, ultra-liberal parsing

["Bob Wyman" <bob@xxxxxxxx>]

Henri Sivonen wrote:
> So I think with Atom we need:
> ...
> 2) Libraries that are so good most people don't 
> want to roll their own.
	This is what you have in the worlds of ASN.1, CORBA, and a
number of other binary protocols already. In those worlds, you would
never imagine that someone would send data that didn't conform to the
rules since typically, everyone uses a small number of well tested
libraries and virtually no one ever actually sees the
bits-on-the-wire. To a great extent, this is also the situation that
exists in SOAP-land. (i.e. the vast majority of SOAP users now appear
to be .NET folk who interface to the wire through layers and layers of
code that ensures that conformance to specifications is maximized.)
	In the XML-world, strict conformance to standards has never
been a "value" of the community. Heck, in recent debates on XLM-DEV,
there were even long treatises being written by people who objected on
apparently "moral" grounds to the very idea that schemas should be
used to describe data or that schemas should be considered
authoritative!
	But, you are right. We've learned with other systems that use
of common components is a marvelously efficient way to get consistency
and interoperation. Fortunately for XML, there are already a number of
such components available. For instance, SAX2 does a wonderful job of
both reading and writing XML and there are a number of alternatives to
SAX2 that are growing in popularity. But, we've still got people who
would prefer to roll-their-own... How do you convince people to stop
playing with the toys and reimplementing things that already exist?

		bob wyman