Lists Home |
Date Index |
Karl Waclawek writes:
> However, one can also make a strong argument against liberal use of
> exceptions for controlling *normal* program flow simply by observing
> how readable and understandable the code becomes, since you now have
> two ways of doing the same thing.
Prematurely terminating parsing is *not* normal from an XML
perspective (it defines both well-formedness and validity in terms of
complete documents), though I agree that there are many software
applications where it can be useful. Given the circumstances (and
insignificant overhead, since it happens only once for each document),
an exception has always seemed like the best fit.
The most obvious alternative would have been to require all callbacks
to return a boolean value, which would be a giant programming pain for
the majority of applications that process the complete XML document.
We could also have set up a method on the parser itself, but in Java
that is much more kludgey than using an exception.
All the best,
David Megginson, email@example.com, http://www.megginson.com/