Re: Notes on the process

Michael Sperberg-McQueen wrote:

> If a life and death application can't rely on parser heuristics to
> recover from errors in the input, then the error recovery should
> probably be in the downstream application, not the XML processor as
> described in the spec, in any case.  So I don't see that the decision
> has made any difference to the plausibility of XML as a choice for such
> applications.

The error recovery should be in the downstream application. The ERB 
decision prevents that by essentially requiring that on error XML processors
stop processing and send along raw data. Recovery in a conforming
application will require processing the raw data without the aid of an XML
processor. 

Many application developers won't bother. If we're lucky, they'll ignore the
error handling language and choose to implement non-conforming applications.
If we're unlucky, they'll choose an alternate representation for their 
information - one that allows reasonable error recovery. 

> But are you willing to bet me your life that the average parser writer
> will correctly guess which well-formed string from among those given
> (and from the infinite number of other well-formed strings that could be
> transformed into the original string by interruptions in the
> transmission) was 'intended' by whatever created the original ill-formed
> example?

I am willing to bet you my life that at some point in the future, there
will be mission critical, fault-tolerant, web applications that have a
need for structured information exchange. Had you asked a similar question
regarding telephone service at the end of the 19th century, many would
not have taken the bet. They would have been wrong. In the US, we bet
our lives on the availability of telephone service and 911 operators every
day.

I don't have the wisdom to predict specific fault-tolerant applications that
will make the Internet an integral part of our lives. I do know that those
applications will require structured information and will not respond to
error conditions with "I'm sorry Dave, I'm afraid I can't do that."

Received on Thursday, 8 May 1997 22:56:46 UTC