Xerces 2.0.2 is now available!

Hi all,

The Xerces-J team is proud to announce that Xerces-J 2.0.2 is now
available.
In this release, numerous bugs have been fixed. Only one minor change i=
n
the Xerces Native Interface is included (documented below); considerabl=
e
new functionality
has also been added in terms of PSVI support, DOM level 3, and grammar
preparsing/caching.

The release can be downloaded from

     http://xml.apache.org/dist/xerces-j/

The major changes made since Xerces 2.0.1 are:

- Added Schema Component API interfaces and implementation; full PSVI
support. [Sandy Gao, Elena Litani]
- Add XMLResourceIdentifier to startExternalSubset() method defined in =
XNI
XMLDTDHandler. That allows supporting baseURI for an external subset.
[Elena
Litani]
- Added implementation for baseURI, documentURI, normalizeDocument and
required normalize document features. Added support for DOM revalidatio=
n
against
XML Schemas. [DOM Level 3] [Elena Litani]
- Changed DOM createElementNS and createAttributeNS methods so that the=
y
now convert the empty string, when used as a namespace URI, to null. Th=
is
is a change in behaviour compared with Xerces 2.0.1, but it is now
permitted by DOM level 2 and it aligns with DOM level 3 expectations.
[Elena Litani]
- Added implementation for compareTreePosition method of DOM L3. [Lisa
Martin]
- Added XMLGrammarLoader interface to the xni.grammars package; refacto=
r
DTD and Schema validation code to create classes implementing this
interface;
created a convenience class (XMLGrammarPreparser) that uses the new gra=
mmar
loaders to permit grammar preparsing; wrote a sample
(xni.XMLGrammarBuilder)
that demonstrates how all this works. [Neil Graham]
- Reorganized the code in Xerces DOM implementation to expose DOM Level=
 3
functionality via org.w3c.dom package. Added several build targets for
building
customized xerces jar files that include DOM Level 3 Core and Load/Save=

interfaces and implementation of those interfaces: jars-dom3, apijar-do=
m3,
sampjar-dom3.
[Arnaud Le Hors, Elena Litani]
- Updated xni.DocumentTracer sample to print augmentations for events. =
This
makes it easier to debug configurations that augment the document's
infoset.
[Andy Clark]
- Removed dependence on SAX attribute interfaces from XMLAttributesImpl=

utility class. The AbstractSAXParser already defines a SAX attribute pr=
oxy
class so that the parser and components can be properly layered. [Andy
Clark]
- Moved from the org.xml.sax.helpers.AttributesImpl shipped with SAX 2.=
0 to
that shipping in 2.0.1. This solves problems encountered with creating
empty attribute lists. [Neil Graham]
- Improved handling of settings in parser classes making it easier to
re-use the parsers with other parser configurations. Previously, the pa=
rser
instances assumed that all configurations supported various settings (i=
.e.
the parser would set default values without first adding the recognized=

features
and property identifiers to the parser configuration). [Andy Clark]
- Fixed bug in HTML DOM implementation that was causing xerces.dom elem=
ent
nodes to be created instead of the appropriate HTML element nodes. [And=
y
Clark]
- Fixed bug to avoid null pointer exception in AbstractDOMParser when
locator information is not available from parser configuration. [Andy
Clark]
- Fixed bug in WrappedInput/OutputStream classes reported by Morten
Bj=F8rhus. [Andy Clark]
The DTD validator failed to report "undeclared element" error if an
attribute was declared for an undeclared element. [Elena Litani]
- Performance fixes: replaced code that increases array sizes by catchi=
ng
ArrayIndexOutOfBoundsExceptions with code that checks the sizes of the
relevant
arrays. Also streamlined several method calls in the DTD Grammar classe=
s to
account for the fact that scope is not relevant for this type of gramma=
r.
Finally,
improved algorithm for recognizing entities and notations. [Henry Zonga=
ro,
Neil Graham]
- Compress jars by default. We have not done so in the past, and it wil=
l be
useful to see if this proves problematic or beneficial. [Neil Graham]
- Added several build targets for building customized xercesImpl jarfil=
es.
Now we can build a DTD-only version, and a DTD-only version also exclud=
ing
support for the HTML/WML DOM. [Neil Graham]
- Parser now reads external entities one chunk at a time, rather than a=

character at a time; this increases performance dramatically in certain=

circumstances.
This fix also allows a number of EBCDIC encodings to work which did not=

work previously. [Neil Graham, Henry Zongaro]
- Performance enhancement: implemented DOM node and declaration pools t=
hat
reduce number of objects created during building of XML Schema grammars=
.
[Elena Litani]
- Improved performance for non-deferred DOM: reduced number of
concatenations for characters (bug #5602) [Elena Litani]
- According to the namespace spec errata, namespace declaration attribu=
tes
should be bound to a special namespace; and errors should be reported o=
n
illegal bindings of "xmlns" and "xml" prefixes and their namespaces. [S=
andy
Gao]
- Refined schema error messages: provide more descriptive error message=
s
for simple type validation errors; avoid cascading errors when a gramma=
r
for
a given namespace is not found, or when components from a certain names=
pace
can't be accessed from a given schema document. [Sandy Gao]
- Changed the lexical representation of gMonth from "--MM--" to "--MM".=

[Sandy Gao]
Decimal point shouldn't be allowed in integer-derived types;
"fractionDigits" and "totalDigits" should be calculated on the value sp=
ace.
[Sandy Gao]
- Fixed several DOM bugs. Xerces now passes W3C DOM Level 1 Core test
suite. [Elena Litani]
- Improved management of ID attributes in the DOM so that
Document.getElementByID is more reliable. [Arnaud Le Hors]
- Applied patch from Scott Sanders to fix Java serialization of the DOM=
 and
add a test. [Arnaud Le Hors]
- Applied patch from Henry Zongaro to fix a ClassCastException in the
deferred DOM. [Elena Litani]
- Applied patch from Fabio Riccardi to reset normalize value in the PSV=
I.
[Elena Litani]
- Applied patch from Joe Kesselman that fixes NPE in a NodeIterator (bu=
g
#6888). [Elena Litani]

Enjoy!
Neil
Neil Graham
XML Parser Development
IBM Toronto Lab
Phone:  905-413-3519, T/L 969-3519
E-mail:  [email protected]
=