>my problem is, whenever i come to a point in the proposed recommendation at
>which a parser is required to report an error and "must not continue normal
>processing" even though the result which the stream would denote would be
>sufficiently unambiguous if allowed, then i feel compelled to ask, "why does one
>have to exclude this"?
The requirement for "Draconian error handling" actually came from the HTML
side not the SGML people. The reason was to ensure data integrity:
if a document was compromised it should be clearly marked as such when
passed to the application. Under no circumstances should something that
is not well-formed be passed to an application as if it were.
This is because XML is intended for more than just typed-text applications.
It was thought that allowing all sorts of transparent error-recovery
mechanisms would just reintroduce tag minimization in through the back
door. Then people would start to rely on it, or at least write their
XML to suit the error-recovery of particular parsers, and we would
be back in HTML-land, where the effective grammar is too loose to
be reliable.
Rick Jelliffe