Lists Home |
Date Index |
- From: Miles Sabin <firstname.lastname@example.org>
- To: email@example.com
- Date: Thu, 13 Jan 2000 14:49:19 -0000
Leigh Dodds wrote,
> Miles Sabin wrote,
> > An XMLReaderFactory could return an Iterator or Enumeration
> > over the available implementation classes (more likely over
> > some meta-information providing access to the features of
> > the various installed parsers so an application can choose
> > if there's more than one) and allow an app to select and
> > instantiate one.
> Why not have the Factory method accept the criteria,
> and return an implementation directly, rather than forcing
> the application to iterate over them?
That'd be another way of doing it. I'm not sure which I prefer.
> In the case where there are several parsers with the same
> properties, it will either matter to the application - in
> which the preference will be hard-coded (not good) as it will
> have to select it from the list, or it won't matter, so
> returning the first match is good enough.
> If there are two parsers with the exact same features its
> likely that there will be other differentiators which can't be
> checked in code. e.g. which is the fastest.
If there are parser features which aren't queriable via the SAX
API then you still have the option of directly instantiating a
particular concrete XMLReader implementation.
> Why not have the equivalent of a BeanInfo file for the parser.
> Just add the package names to the XMLReaderFactory's property
> files and the properties can be checked at runtime. This
> seems less complex. Loading and parsing a file seems like more
> overhead than instantiating a BeanInfo equivalent and querying
> it directly.
BeanInfo strikes me as a bit heavyweight. Maybe my description
made the proposal sound a bit more complex than it really is.
For most parser implementors all that would be involved would
be sticking the fully qualified class name of their parser
into a text file located at org/xml/sax2/XMLReader.providers.
As to parsing, it's not a lot more than reading a line of text
from a text file and invoking Class.forName().
> Anyway, doesn't the JAXP package address some of these issues?
I don't see that this is very different from the current scheme
in SAX1. Sure, they've added another layer of indirection: the
javax.xml.parsers.SAXParserFactory System property names a
factory implementation class rather than a parser implementation
class. But you'd still have to set the property in the
environment, and you still wouldn't be able to use that
mechanism to bind to more than one parser.
Hmm ... I guess you'd also have to implement their
SAXParserFactory interface too: nothing like that is necessary
in SAX1 nor would it be under my proposal.
Miles Sabin Cromwell Media
Internet Systems Architect 5/6 Glenthorne Mews
+44 (0)20 8817 4030 London, W6 0LJ, England
xml-dev: A list for W3C XML Developers. To post, mailto:firstname.lastname@example.org
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/ or CD-ROM/ISBN 981-02-3594-1
Please note: New list subscriptions now closed in preparation for transfer to OASIS.