[Xerces2] Workshop Report

[Ted Leung <tw...@sauria.com>]

Hi folks,

Below is a tardy summary of the Xerces2 Workshop that was held this
past Monday


The Xerces-2 developers conducted a workshop on Xerces 2 in Cupertino
on Monday June 11.  There was pretty reasonable turnout from the main
Xerces-2 developers and interested parties.

Much of the workshop was devoted to presentations about the current state
of the Xerces 2 implementation along with discussions of various issues that
need to be resolved in order for development to continue.  As Andy has
already posted, these slides are available at
http://www.apache.org/~andyc/speeches/20010611/xerces2ws.ppt

Kohsuke Kawaguchi summarized his experience porting his SAX based
Relax verifier to XNI:
   1. Documentation would have helped
   2. He was impresssed with how easy (relatively it was to do)

Here is a list of items that needs to be completed in no particular order:

1. We need a better way to update the website - this is making it hard for
the
Xerces 2 developers to get the word out.  This is a general problem with
xml.apache.org
2. We need better documentation for Xerces 2 - is there someone out there
who would like to get involved with documentation?
3. We need to finalize the XNI apis.  In order for this to happen, the
developers would
like to get some feedback from people who attempt to use the XNI apis.  To
date a
single XNI component has been implemented.   We could really use feedback
from
Xalan and Axis in this area.   Below is a list of the big open areas in XNI.
    a. Validation - one of the big open areas is the redesign of the
validation engine to
        switch from on the way out to on the way in validation.  Kohsuke
Kawaguchi suggested
        looking at things from a different angle.  The current on the way
out validator builds an
        automata checker for each element's content model and glues them
together using Java
        flow control.  Kohsuke suggested that instead we build an automata
for the entire grammar
        at once and let the state machine handle everything.   This approach
seems to have a lot of
        benefits -- it retains the good performance of the on the way out
scheme while allowing us
        to do a better job of reporting errors than we do now, as well as
solving the problems with
        <any> in XML Schema.  There are a few open issues with this approach
including:
          i.      handling of xsi:type
         ii.      handling of complex type inheritance
        iii.      modularity issue w/ schemaLocation on non-root
lements.  - this might be overcome
                  by lazily creating parts of the automaton.
    b. XMLDTDHandler - 1 or 2 interfaces
    c. XMLString or String or Stringpool & consistency of their use
4. Before Xerces 2 could go into widespread use it needs to:
    a. Pass the entire OASIS XML conformance suite - validation in
particular
    b. Include support for XML Schema
5. We need to create list of things that people could help out with.

Ted


---------------------------------------------------------------------
In case of troubles, e-mail:     webmaster@xml.apache.org
To unsubscribe, e-mail:          general-unsubscribe@xml.apache.org
For additional commands, e-mail: general-help@xml.apache.org