[
Lists Home |
Date Index |
Thread Index
]
On 5/6/05, Chris Burdess <dog@bluezoo.org> wrote:
> Peter Hunsberger wrote:
> > Just out of interest: when are "users" going to be switching between
> > parsers for a given application? Even as a developer once I've picked
> > a particular parser for a particular application I've pretty well
> > committed to it for the lifetime of the app...
>
> In Java, the standard means of retrieving a parser implementation (JAXP)
> is user-overrideable. The user can choose a different parser by
> specifying different system properties. They may decide to run your Java
> application on a different platform, one for which the default parser is
> different. They may do this to experiment with different configurations
> and performance and stability, or for instrumentation, or they may be
> completely unaware of it, the decision to use another implementation
> having been made by someone else (following an operating system upgrade,
> etc).
>
> If you're used to developing in C with libxml2, this may sound arcane.
> But even within such a static codebase, software changes: new project
> managers and system architects come and go, and refactor the "legacy"
> code according to their own experience and understanding of the domain,
> including the XML specification.
I develop with Java, I'm well aware of JAXP and its kin. They're nice
in theory, but for any moderate sized application, even with "100%
conforming" parsers (whatever that might mean), any developer who
leaves it to chance as to what parser is picked at any given moment is
just asking for trouble; if you're not careful you might fall back to
the default Xalan parser (bugs and all ;-)
--
Peter Hunsberger
|