Todo:

1.4:
---

digital signatures?
encryption?

1.3:
-----

DTD API?

catalog support????

RELAX NG?

add an XPath based NodeFactory that only instantiates nodes that
satisfy a certain pattern

A util package with the most useful samples and contribs

Do Serializer and canonicalizer need setOutputStream methods?

should Builder have an option to cache all external entities so it doesn't keep reloading them, perhaps via a SAX EntityResolver.

A getAttributeValueInScope() method that searches up the tree for the nearest 
ancestor element with the specified attribute. This would be useful for xml:lang, xml:space, and many other cases!!!!

Should Builder/NodeFactory have a build(Document) method that allows one to pass an existing document through a node factory without reserializing? 

Should/could the TrAX source and result, Jaxen adapter and so forth go public in the converters package?

1.2:
-----

Look at some more of Wolfgang's optimizations

could some form of filterlist improve xpath performance?

should there be a special iterator for *; can this be combined with named iterator?

should //* somehow avoid the resort?

could we simplify /descendant-or-self::node()/child::* as 
/descendant-or-self::*/child::* | /self::node()/child::*

Optimize normalization

test running in unsigned java web start

add a NodeFactory section to the tutorial for processing big documents

check if we've accidentally sealed in jaxen
what should the manifest say about that?

Make a build target for non-LGPL, closed source version

Sign up on Kagi or somewhere for software sales

Add a Buy XOM option to the main XOM sidebar, the unstable sidebar. and the
license page

Profile AllTests (or maybe FastTests). 

Add Class-Path entries to manifest

Add more XPath samples based on XOM

Figure out a way to allow users to configure different text storage algorithms,
all stored as byte arrays.
Possibilities:

System property (though that's VM wide, which is really, really bad)
Method in text class and Builder

StAX Builder and converter sample to test API and implementation

Add separate implementations optimized for speed and size.
The current implementation is mostly optimized for size. The speedy version
would carry extra pointers (especially next sibling) to make navigation very quick.
Some final methods would have to be implemented by calling non-final package protected 
methods that could be overridden instead.

Should XSLTransform have a constructor that takes a TrAX 
Templates or Transformer object to allow
additional properties to be configured

A FastSerializer that does no indenting, no line-breaking, no character sets
except UTF-8 and UTF-16, no normalization, and does not use TextWriter

consider using a WeakHashMap to hold all mappings of nodes to base URIs

Could Builder hold a cache of interned strings to use while building?
Does NUX BinaryXMLCODEC do something like this?

Add a binary distribution that's like the full distribution but no source.
Should the source distribution not include the binary?

Could DOMConversion be done in parallel in two separate threads?

DHollenbeck suggests using char[] arrays internally rather than Strings
 would save memory and time.
See http://lists.ibiblio.org/pipermail/xom-interest/2004-January/000898.html

Can XInclude be rewritten to be non-recursive? Isn't it already?

with XIncluder would it make more sense to use a private class
that stores all the variables rather than constantly passing them
back and forth from static methods? That is, make private methods instance 
methods rather than static????

would it be possible to get a minimal interoperable subset of Big5, SJIS, etc
and escape all non-interoperable characters?

A streaming serializer based on a  NodeFactory that can serialize arbitrarily large documents without having the entire document in memory

should I synchronize a single static TransformerFactory object inside XSLTransform?

Would it make sense to cache previously verified namespace prefixes 
to avoid checking them every element and attribute? 

Ditto, would it make sense to hash previously looked up names?
Must profile this. 

Add a Fetch task that loads code from CVS
(Need to fix guest access on java.net first, or register a fake user)

Use GET task to grab servlet.jar and tagsoup.jar if necessary

XInclude:  In order to minimize encoding errors for parse="text" processing,
please change the definition of the encoding attribute to include a
requirement that if the attribute has a legal value and the encoding is
supported and the protocol supports such action, that the server is
informed of the encoding attribute value, e.g. for encoding="iso-8859-2"
and a HTTP request, that the request includes

  Accept-Charset: iso-8859-2

such that the server has a chance to provide a proper representation.

would it make since to start with a larger size for attributes and children when building and trimToSize when done?

Clean up classpaths in build.xml.

could I set up JDK15Parser to compile as individual tasks
that are only compiled when dependencies are satisfied?
For 1.5 use ant.java.version property and an equals condition

Wolfgang's ant build file for smaller vs. faster jars (Text class)

Add tools page to web site (Ant, TagSoup, Clover, etc.)

cvs -d :pserver:elharo@cvs.dev.java.net:/cvs rtag XOM_11a3 xom


Backport Fixes to make in 1.0.1
-------

TextWriter.java 1.1a2
XOMTestCase.java
XOMHandler.java for internal DTD subset
Serializer for NFC in consecutive text nodes
XIncluder for base URLs and better exception messages
XOMReader and SAXConverter to allow XSL to work with xml:base attributes
XOMHandler: Workaround for Crimson bug that fails to report asterisk after mixed content declarations
XOMHandler: Workaround for Crimson bug that makes xmlsn:xml="correct URI" and error
XOMHandler: Workaround for Crimson external and internal DTD subset mixup bugs

Hi Elliote.
http://www.cafeconleche.org/XOM/apidocs/nu/xom/xslt/XSLTransform.html

has an example

public static Nodes transform(Document in) 
           throws XSLException, ParsingException, IOException {
             Builder builder = new Builder();
             Document stylesheet = builder.build("mystylesheet.xsl");
             XSLTransform stylesheet = new XSLTransform(stylesheet);
                           ^^^^^^^^^
             return stylesheet.transform(doc);
         } 

stylesheet used twice !






~/projects/1.0.1$ cvs -d :pserver:elharo@cvs.dev.java.net:/cvs rtag -r XOM_10  -b BR_1_0 xom
lock.c:222: failed assertion `strncmp (repository, current_parsed_root->directory, strlen (current_parsed_root->directory)) == 0'
cvs [rtag aborted]: received abort signal
cvs [rtag aborted]: received abort signal
lock.c:222: failed assertion `strncmp (repository, current_parsed_root->directory, strlen (current_parsed_root->directory)) == 0'

===============
Done 1.2b1
===============

xml:id attributes no longer checked for NCNames

Upgraded to Xerces 2.8.0, DTD-only version

DOMConverter can accept a NodeFactory to be used in creating the XOM document

No longer possible to set an attribute's type to null.

Jaxen source is bundled. Ant no longer checks it out of CVS.

Added a lookup method to XPathContext that retrieves a namespace URI given a prefix


===============
Done 1.1
===============

Documentation updates

===============
Done 1.1b7/RC1:
===============

Fixed bug that could unnecessarily escape carriage returns and linefeeds and numeric character references

Fixed bug that could sometimes change line breaks

Avoid leaking memory from Builkder when not reusing it.

===============
Done 1.1b6:
===============

Fixed bug that could append a text node to a document when parsing a malformed document, followed by a well-formed document. 

Fixed infinite loop in Canonicalizer when canonicalizing an element with at least two ancestors but in no document

Canonicalizer no longer puts detached elements in a document when canonicalizing them


===============
Done 1.1b5:
===============

Small optimizations in Attribute class

Fixed bug in Canonicalizer when canonicalizing a non-detached element


===============
Done 1.1b4:
===============

Fixed bug in SAXConverter where start/endNamespacePrefixMapping could be called multiple times for the same namespace


===============
Done 1.1b3:
===============

Added SUIDs to Serializable classes (mostly exceptions)

Numerous optimizations including:

Replacing several stacks with ArrayLists
Using an unsynchronized custom BufferedWriter for serialization
Using String instead of StringBuffer in characters() in XOMHandler


CDATASection.toXML() now outputs its value wrapped in a CDATA section rather than escaped.

Bundling Xalan 2.7 instead of 2.6 and Xerces 2.7.1 

Various workarounds for bugs in Xalan 2.7 and changes in Xerces 2.7


===============
Done 1.1b2:
===============

Child nodes and attributes are now stored directly in arrays without an intermediary list.
This saves some memory.

A few bug fixes, especially in XPath


===============
Done 1.1b1:
===============

Lots of Jaxen fixes

Lots of Crimson fixes

===============
Done 1.1a3:
===============

Normalization form C serialization is now correct
even when the characters that need to be combined cross the
boundaries of consecutive text nodes

In addition, Serializer does its own normalization. There is no longer any dependence on ICU.

XInclude sometimes generates relative URLs when doing base URI fixup.

XSLT can now operate on xml:base attributes

===============
Done 1.1a2:
===============

XPath is an order of magnitude faster

XPathTypeException class

XPathDriver sample program

Small bug fix in XOMTestCase

Various bug fixes in Jaxen

===============
Done 1.1a1:
===============

A few small speed ups

Some bug fixes in XOMTestCase

===============
Done 1.1d6:
===============

The primary JAR file now bundles Jaxen so that in Java 1.4 and later the only
thing that needs to be on the classpath is xom-1.1d6.jar (provided you don't use NFC in the serializer)

A setParameter method in XSLTransform

All items in a Nodes must now be non-null

XPath expressions now recognize the xml: prefix, even if it hasn't been
specifically bound in a context

Fixed bug that unnecessarily duplicated xml: attributes 
during document subset canonicalization

The Canonicalizer API has been reworked significantly. 
Document subset canoniclization is now performed by passing in a Nodes
object rather than a Document and an XPath expression. The Canonicalizer(out, boolean, boolean) constructor
has been removed on the grounds of redundancy and confusion. The canonicalize(Document) method is now canonicalize(node) and canonicalizes the entire subtree represented by the ndoe you pass to it.

===============
Done 1.1d5:
===============

Exclusive XML canonicalization

===============
Done 1.1d4:
===============

XPath expressions can now return Namespace nodes

Document subset canonicalization

Fixed bug that prevented round tripping of \r\n in attribute values

===============
Done 1.1d3:
===============

xml:id support

===============
Done 1.1d2:
===============

XPath support

Preserve all entity declarations in internal DTD subset
because these may be needed by the external DTD subset

Bugs Fixed:
-----------

Escape percent signs in the internal DTD subset
to prevent accidental interpretation as parameter entity references

Workaround for Crimson bug that does not use parentheses when reporting NOTATION names for attributes

Can now handle " and &in the internal entity declarations in the internal DTD subset

Better handling of weird filters that skip expected steps like startDocument

XOMTestCase.compare(Node, Node) throws ComparisonFailure when comparing nodes of different types. 

XOMTestCase.compare(Node, Node) throws ComparisonFailure when comparing nodes of different types. 

XOMTestCase.compare(Node, Node) can now compare attributes

Parses better with parentless filters

===============
Done 1.1d1:
===============

setInternalDTDSubset

===============
Done 1.0:
===============

Update all versions to 1.0

Add an example of running canoniucalizer and/or prettyprinter to README

===============
Done 1.0b11/RC5:
===============

Servlet samples restored, but build file only
compiles them if servlet.jar is present

LGPL, license, and readme files added to distro

===============
Done 1.0b8/RC2:
===============

The TagSoup and servlet JARs are no longer bundled. They're not needed to run XOM, just for one of the samples and for the JavaDoc

A few more optimizations to speed up the checking of namespace URIs, and a variety of other operations.

===============
Done 1.0b7:
===============

Comments whose data begins with a hyphen are now allowed. 

Builder is considerably more robust against buggy parsers. It converts all
runtime exceptions thrown by such a parser 
(including XOM XMLExceptions thrown by a NodeFactory) 
into ParsingExceptions.
It uses a verifying factory for Saxon 7's AElfred derivative.

XIncluder treats bad encoding attributes as fatal errors

Various optimizations have sped up a lot of common operations
including getValue(), toXML(), DOM and SAX conversion, canonicalization,
and XSL transformation

The zip archives and CVS no longer contain files with names that are problematic on Windows.

The manifest file is now versioned.

In keeping with the recommendation in RFC2396bis that "For consistency, URI producers and normalizers should use uppercase hexadecimal digits for all percent-encodings", XOM now uses uppercase percent encodings for base URIs. There may still be a few places where lower case escapes are used. 
Holler if you spot any. 

Fixed bug where base URIs were not encoded in UTF-8 on all platforms. Mac OS X 10.3 was the particular offender here. Surprisingly the problem did not manifest on Mac OS X 10.2. 

===============
Done 1.0b6:
===============

SAXConverter no longer converts XOM xml:base attributes into SAX attributes.
Instead the xml:base attributes are used to determine the URI information
the locator reports. Providing xml:base attributes as well would risk 
double counting some relative URLs. 

Fixed a number of bugs in converting file names to base URIs 

Improved compatibility with Turkish locales that do not see I as the 
upper case form of i and vice versa

Fixed bug where carriage returns in internal entity 
replacement text in the internal DTD subset was not properly escaped
on reserialization

Fixed bug where carriage returns, less than signs, double
quotes, and ampersands in attribute default values in the 
internal DTD subset were not properly escaped
on reserialization

Hid the error messages logged by Xerces and Xalan on System.err
when deliberately testing error conditions. Therefore, there should
be no output fropm the test cases when all tests pass.

Added a junithtml build target to convert JUnit results to HTML. 

The strings returned by toString in Comment, ProcessingInstruction,
Attribute, and text are all now truncated if they get too long.
Furthermore any embedded line breaks and tabs are escaped as \n, \t, 
and \r. This makes the objects easier to inspect in various debuggers.

The Ant build file now specifies that the input encoding of all .java files 
is UTF-8. Most files are pure ASCII, but there are a couple of places where 
non-ASCII characters are used. 

Unit test coverage has been improved slightly.

Fixed a bug in Serializer that did not always properly trim whitespace

===============
Done 1.0b5:
===============

XSLTransform.setNodeFactory is deprecated. Instead use 
the new XSLTransform(Document, NodeFactory) constructor.

XIncluder now resolves XPointers in xi:include elements 
against the acquired infoset rather
than the source infoset.

Scheme specific errors in XPointers are treated as resource errors when
xincluding rather than fatal errors.

Much faster when building documents from File objects.

Deprecated constructors have been removed from XSLTransform

===============
Done 1.0b4:
===============

XSLT transformation is now based on SAX conversion rather than toXML.
This should save memory in transformations and probably speed things up.

All constructors in XSLTransform that take anything other than a Document are deprecated
and will be removed in the next release:

public XSLTransform(InputStream stylesheet)
public XSLTransform(Reader stylesheet)
public XSLTransform(String URL)
public XSLTransform(File stylesheet)

SAXConverter can now convert Nodes lists as well as Documents.

SAXConverter now provides location information for system IDs.

Various bug fixes in SAXConverter, especially with respect to startPrefixMapping 
and endPrefixMapping

The toXML methods now use \n as the line separator, since this is more likely
to match the contents of text nodes created by parsing an XML document.
The goal is to minimize the number of documents with mixed line break strings.

DOMConverter can now convert XOM documents with only a single element to DOM.

Minor bug fixes to better handle line breaks in the internal DTD subset

===============
Done 1.0b3:
===============

Java encoding names are now recognized when using the repackaged Xerces
bundled with Java 1.5

Fixed several bugs in DOMConverter

===============
Done 1.0b2:
===============

Fixed various bugs that prevented the loading of JDK15_XML1_0Parser

Worked around bugs in JDK 1.5 beta 2 that limited elements to a single attribute.

The API documentation is now well-formed XHTML. (It might even be valid. I haven't checked.)

There's a new tools package that contains classes used to help build XOM. Currently this 
contains the class to convert JavaDoc to XHTML using TagSoup.

===============
Done 1.0b1:
===============

The XInclude test suite is loaded and run from the W3C CVS 
server if necessary

Worked around various JDK bugs that prevent round-tripping of 
some characters in Japanese encodings

Improved compatibility with Java 1.5

===============
Done 1.0a5:
===============

ParsingException and ValidityException now supply the URI of the document that caused 
the exception if it's available

OASIS XSLT conformance tests are now included in the unit test suite

Handling of additional namespaces in transforms now works with
recent versions of Xalan

Improved compatibility with pre-1.4 VMs

===============
Done 1.0a4:
===============

Nodes.remove(int) now returns the node removed

The IBM virtual machine 1.4.1 is no longer special cased.

The API documentation has undergone extensive editing. 

The unpublished nu.xom.xerces package has been removed.

===============
Done 1.0a3:
===============

The Element copy constructor and copy methods are no longer recursive, so they
shouldn't cause stack overflows in deep documents. This necessitated adding a 
protected shallowCopy() method that can be used to create an instance of a subclass
of Element. Overriding this is preferred to overriding copy() when one wishes 
to maintain the objects' types after a copy.

The getBaseURI() method is also no longer recursive. 

The W3C XML Schema Language and WML and HTML DOMs have been removed from 
the bundled version of Xerces to save space. 

There is now a contributor license agreement.

XOM will now use character references only when necessary for
*all* encodings supported by the local virtual machine.
However, this may be quite a bit slower than the 
explicitly supported encodings like UTF-8 and the ISO-8859
character sets. Measurements remain to be performed.

===============
Done 1.0a2:
===============

URI verification and base URI resolution are now performed 
according to the RFC2396bis algorithm, rather than by using the 
Xerces and java.net URI classes.

The Builder no longer sets any System properties for more compatibility
with applets and multiclassloader environments.

Fixed bug in DOMConverter


===============
Done 1.0a1:
===============

The base URI handling has been modified as follows:

1. getBaseURI() always returns an absolute URI or the empty string if the base URI is not known.
   Other than the empty string it never returns a relative URI. 
   It never returns null.
2. Base URI of an element does not change when it is detached or copied
3. setBaseURI requires an absolute URI, and throws a MalformedURIException if you attempt
   to pass it a relative URI, or a URI with a fragment identifier. (Relative URIs are still allowed in 
   xml:base attributes.)


XOM will not double verify when being fed data through Norm Walsh's catalog filter;
provided that the underlying parser is good.

Supports 2nd candidate recommendation syntax for XInclude

Constraints on parentage are not checked when building with Nonverifying factory
(fastAddAttribute, fastInsertChild) 

DOMConverter is now non-recursive

An element's absolutized base URI is preserved when detaching

===============
Done 1.0d25:
===============

The checkFoo methods have been eliminated. All
setter and mutator methods in the node classes are now non-final.

NodeFactory.makeDocument has been renamed startMakingDocument
NodeFactory.endDocument has been renamed finishMakingDocument

Added a method to DOMConverter to convert a DocumentFragment to a Nodes

Added XSLTransform.toDocument() method that converts a Nodes to
a Document

Added UnavailableCharacterException, a subclass of XMLException,
 to be thrown when attempting to serialize a
character that is not available in the current charater set and cannot be escaped

Element.addAttribute is declared to throw the more specific MultipleParentException instead of 
IllegalAddException

Added a non-recursive serializer sample

Removed checkDetach() method from Node. It was redundant with
checkRemoveChild() in ParentNode.

Reursion has been eliminated from several methods in Element
to make it work better in very deep documents; notably
toXML(), getValue(), and getNamespaceURI(prefix)

The canonicalizer has been made non-recursive

ParentNode.replaceChild() will not remove the old child unless it can insert 
the new child. It can no longer do one but not the other.

Document.replaceChild now allows replacing of the DocType by another DocType
or the root element by another element

Element.removeChildren() now either removes all children or none.
It also returns a Nodes object containing the children removed

LeafNode has been removed. DocType, Text, Comment, and ProcessingINstrcution
now directly extend Node.

Removed hasChildren method from Element, Node, ParentNode, and Document

Much better testing of canonicalizer. I am now fairly convinced
it is correct in all or almost all cases. 

Line breaks are now used between declarations in internal DTD subset

Compiled jar without debugging symbols to save space.
(These can be turned on again easily enough in build.xml
if anyone needs them.)

Made a XOMSamples.jar

The core JAR archive is sealed

Many JavaDoc improvements


===============
Done 1.0d24:
===============

Fixed resource loading in servlet/multiclassloader environment

===============
Done 1.0d23:
===============

Added support for accept, accept-charset, and accept-language
attributes on include elements

MissingHrefException has been renamed  NoIncludeLocationException

XOMTestCase is part of the published API.

CircularInclusionException has been renamed InclusionLoopException

Factory methods are now invoked in document order. Previously this
wasn't true for text nodes, which weren't flushed until after 
the next tag, PI, etc. This was necessary to enable text nodes to
be maximally contiguous, though in fact they might not be if
the factory returned several text nodes in a row for non-text nodes.
In any case, with the default factory, or with a custom factory that
doesnot remove any nodes or change their base types (e.g. coment to Text)
text nodes are still maximum contiguous after a build.

Added support for GB18030 encoding on output
(requires Java 1.4)

IllegalDataException and its subclasses have getData and setData methods
to get and set the exact text that caused the exception.
IllegalNameException, IllegalCharacterDataException, and IllegalTargetException are now 
subclasses of IllegalDataException. IllegalCharacterDataException replaces most previous
uses of IllegalDataException.

NamespaceException has been subdivided into 
IllegalNameException, MalformedURIException, and NamespaceConflictException. 

Verifier is now based on table lookup. 

XOM no longer contains any JDOM code.

Removed NodeFactory makeWhiteSpaceInElementContent() method

Serialization speed-ups for Non-Unicode, non-Latin-1
encodings

It is now possible to supply a NodeFactory to XSLTransform to be used for
construcing nodes in the result tree

Improved support for IBM JVM 1.4.1

Added support for Thai in ISO-8859-11/TIS-620 encoding

Speeded up Serializer for non-Unicode/non-Latin-1 encodings

Attribute.Type.toXML is now Attribute.Type.getName(). This was necessary
to be consistent with handling attributes of type ENUMERATION, which is not a DTD keyword
though it is referenced in the Infoset.

Removed no-args constructors from the various exception classes.

The Nodes class now has insert and remove methods,
in addition to append. 

Supports the XInclude 2003 2nd last call working draft.
The methods that resolve Nodes objects have been marked private.

Added NoSuchAttributeException for parallelism with NoSuchChildException

Unit tests have been dramatically expanded. There are now over
700 separate test methods, many of which perform several tests.

No longer allow the namespace URI
http://www.w3.org/XML/1998/namespace
to have any prefix other than xml, per conformance with
the namespaces erratum

Allow the xml: prefix (with the right URI) to be used on elements
per conformance with the namespaces recommendation

NodeFactory make methods now return Nodes objects that may change the type
or number of nodes returned, subject to the ususal XML well-formedness constraints.

Better exception messages when name and namespace arguments are swapped

getBaseURI returns null if the base URI can't be determined due
to a malformed xml:base attribute. 

===============
Done 1.0d22:
===============

Serializer.preservebaseURI() is now Serializer.setPreserveBaseURI()

Carriage returns are no longer allowed in comment and processing instruction data
because they can't be roundtripped. (Character references aren't resolved inside
comment and processing instruction data.)

Initial white space is not longer allowed in processing instruction 
data because this cannot be roundtripped.

DOMConverter.translate methods have been renamed DOMConverter.convert

DOMConverter can now convert individual DOM nodes into XOM objects.
It is no longer limited to converting entire documents.

ValidityException now has a getDocument() method which returns the
complete well-formed but invalid document. It also has getValidityError(int n),
getLineNumber(int n), and getColumnNumber(int n) methods which return 
information about the successive validity errors in the document.

Numeric character references now use upper case.

In Serializer, writeMarkup has been renamed writeRaw and writeText 
has been renamed writeEscaped
since in subclasses these may not actually be writing markup

Much more fine-grained control of serialization from subclasses
using several new methods including writeXMLDeclaration(),
writeStartTag(), and writeEmptyElementTag().

Added an option to serialize using Unicode normalization form C.

Added a protected getColumnNumber() method to Serializer to assist subclasses that
wish to do implement their own line breaking strategies.

Can now specify a Builder to be used when XIncluding

More XPointer syntax errors are detected when XIncluding

NodeList has been renamed Nodes.

Java encoding names such as ISO8859_1 are now recognized on input
if Xerces is the parser. 

XIncludeException (and its subclasses) can now report the URI
of the document where the problem was detected

Upgraded to Xerces 2.6 nightly build to fix bug
involving relative URL resolution in documents 
loaded from redirected URLs

Added unit tests for SAXConverter

Added DatabaseBuilder sample based on Example 8-13 from Processing XML with Java

Silently preserve CDATA sections from parse to output when possible,

Added SourceCodeGenerator sample program that converts a well-formed XML
document into the XOM statements necessary to create the document

Renamed ParseException to ParsingException

===============
Done 1.0d21:
===============

Added checkDetach protected method in Node. Could this 
and checkRemoveChild in Document make code any simpler by preventing
detaching of root?

copy() method is no loinger final in node classes

Cycles (an element acting as its own ancestor)
are no longer allowed. Attempting to create one throws a
CycleException.

NodeFactory.makeDocument() no longer takes an Element as an argument.
It is the responsibility of the NodeFactory to construct a suitable root 
element. However, when parsing this will quickly be replaced by the 
actual root element.

Serializer.setIndent throws an IllegalArgumentException
for negative values

Fixed bug where line breaks would be added if indenting, even in elements
where xml:space="preserve"

XInclude now consistently treats XPointers that don't match any
subresource as resource errors, rather than including nothing.

xml:base attributes added to XIncluded elements no longer 
have fragment IDs

A couple more XPointer syntax errors are now detected when XIncluding

In XIncludeException the getRootCause and setRootCause() methods have
been replaced by initCause() and getCause().

The initCause method in the various exception classes 
is now much more consistent with its definition in Java 1.4.

XSLException no longer extends XMLException. This means it is now a checked
exception instead of a runtime exception.

Xalan 2.5.1 has replaced Saxon 6.5.2 as the bundled XSLT processor
due to a bug in SAXON that incorrectly reported document fragments 
resulting from XSL transforms 

Minor usability improvements and code cleanups in the build.xml file

Added an overview page to the API docs

Cleaned up the API docs, especially for the exception classes

===============
Done 1.0d20:
===============

build test now excludes MegaTest and XOMTestCase

Do not compress jar archive in order to load classes faster

build(File) now works properly on Windows. This fixes numerous
unit test failures in Windows


===============
Done 1.0d19:
===============

By default, the serializer and toXML methods now use
numeric character references to
to escape all tabs, carriage returns, and line feeds in 
attribute values and  carriage returns in text nodes. 
This helps make round tripping more reliable and robust.
However, if the user indicates that white space is not significant
by calling either setMaxLength or setIndent, then these characters
will not be preserved. If the user calls setLineSeparator,
then tabs will still be preserved but carriage returns and line
feeds may not be. 

Cleaned up unit tests

Major speed improvement in the Node.equals() method.
It now executes in about half the time.

For symmetry makeElement is now startMakingElement
endElement is now finishMakingElement.


Characters from Planes 1 to 15 are now escaped correctly 
by the serializer

===============
Done 1.0d18:
===============

Made XOMTestCase public, and cleaned up its code and API
for greater consistency with junit.framework.TestCase

Added support for streaming and partial builds by subclassing 
NodeFactory

===============
Done 1.0d17:
===============

XSLTransform is final

Added unit tests for toString methods and fixed various bugs 
thereby uncovered

IPv6 URIs of the form described in RFC 2732 are now allowed

Fixed various bugs in XInclude. It can now process all the test cases that
do not use the xpointer() scheme or unparsed entities.

The correct exception is now thrown when validating with Crimson.

You can now build with Crimson in Java 1.4.1

Removed numerous unused local variables thanks to pmd

Removed some duplicate code in Builder and Verifier thanks to Same

===============
Done 1.0d16:
===============

The standard jar file no longer includes the samples, tests, and benchmarks packages

Moved SAXConverter and DOMConverter out of the core package into a new 
nu.xom.converters package

Improved compatibility with Java 1.2

SAX filters can no longer bypass well-formedness checks

Worked around a Xerces and Crimson bug that inhibits relative URL
resolution from pathless base URLs such as 
http://www.cafeconleche.org 

The FibonacciSOAPClient sample program works now

More accurate exception messages from the XSLTransform constructors

XSLT unit tests

The distribution now includes the SAXON jar archive so that XSLT works with 
Java 1.3.

Fixed a nasty bug in Element.toXML that was making XSLT transforms
fail when elements were in the default namespace

You can now transform a NodeList as well as a complete document

Document.insertChild(DocType, position) now throws an IllegalAddException
if the Document already has a DocType, rather than silently replacing it.

===============
Done 1.0d15:
===============

Serializer no longer wraps and indents text when xml:space="preserve", 
regardless of the setting of indents and maxlength

XIncluder now has unit tests.

XIncluder now adds xml:base attributes to included elements as necessary

Fixed bug in Document and Element copy constructors that failed to preserve base URI.
This also fixes a bug in the XIncluder

The Element.getChildElements(String name, String namespaceURI method)
now allows a null or empty string local name to stand for any local name,
so you can use this method to get all elements in a certain namespace.
The null namespace is auto-converted to the empty string namespace.

===============
Done 1.0d14:
===============

Worked around some bugs in Xerces that caused the wrong exception
to be thrown when validating.

Improved compatibility with Crimson, the default parser
in Java 1.4

DOMConverter unit tests no longer depend on Xerces

insertChild(String, int) now throws a NullPointerException if String is null

Fixed a nasty bug in Document's copy constructor

Disallowed fragment identifiers in system literal URI references

Fixed several bugs involving the handling of
 notation and unparsed entity declarations in
the internal DTD subset

Added unit tests for internal DTD subset

Fixed build.xml to point at Xerces properly

NonVerifyingFactory for use with parser-created documents

Text class stores content in UTF-8 internally. This reduces memory usage but increases
execution time.

===============
Done 1.0d13:
===============

Several additional methods in Element were marked final (getAttributeCount(),
getNamespacePrefix(int index), removeChildren(),
and getAttribute(int))

Big memory optimizations

The arguments to insertChild (and checkInsertChild and checkRemoveChild) have been reversed.

removeChild now returns the Node it removes

Added support for EBCDIC-37

Fixed another bug on changing namespace URI in XHTML documents

If the Serializer's line separator is set, then all line separators are
changed to that separator on output. If the line separator is not explicitly
set, then all line breaks in source text are preserved as is.

The Builder method
    public Document build(String document, String baseURI) 
now throws an IOException in the event that an IOException occurs while
parsing the external DTD subset
    
Improved unit testing for the serializer and Builder

Removed the equals() and hashCode() methods from the XSLTransform class.
They're probably not necessary, and their behavior was underspecified. 

The public and system IDs of DocType can now be the empty string,
in conformance with the XML spec.
    
===============
Done 1.0d12:
===============

Removed insertBefore and insertAfter from ParentNode

Fixed a bug on changing namespace URI to empty string on 
an element that has attributes

===============
Done 1.0d11:
===============

Added an ANT build.xml file

Wrote unit test for all 6 build methods in Builder and for preserveBaseURI
in Serializer

Worked around bug in later versions of Xerces that doesn't like 
null entityresolvers

Allow base URIs that contain % escapes

Fixed bug that throws NullPointerException when serializing documents without a base URI
with preserveBaseURI set

===============
Done 1.0d10:
===============

Fixed namespace handling

===============
Done 1.0d9:
===============

Removed vestigial getNextSibling() and getPreviousSibling() methods from Document

In Comment:
   Renamed check to checkValue
   Renamed setData to setValue

In ProcessingInstruction:
   Renamed checkData to checkValue
   Renamed setData to setValue

In Text:
   Renamed check to checkValue
   Renamed setData to setValue

In ParentNode:
   Renamed checkRemove to checkRemoveChild for symmetry with checkInsertChild 
   Moved thee two methods down into Element:
   
     public final void appendChild(String text) {

Fixed Builder bug that prevented parsing File objects whose
filenames contained spaces and other non-URL legal characters 

Fixed equals() method in Attribute.Type to work in mutliclassloader
environments 

Corrected usage instructions in samples programs to include the package name
 
Added checks on values of xml:base attributes that they are legal IRIs.
Mainly this involves checking the hex escaping.
   
===================
Done 1.0d8:
===============

XSLT works modulo some obscure bugs in handling the
undeclaration of the default namespace. I need to get some 
clarification on the proper behavior of SAX processors to 
fix this. The TrAX XOMSource and XOMResult classes are not yet
public because I'm still thinking aobut the proper API for these,
but you can use the XSLTransform class for most use-cases. 

It is now possible to undeclare the default namespace on a prefixed element
by passing the empty string as the prefix and URI to declareNamespace().


===================
Done 1.0d7:
===============

Added constraint that an element cannot have two attributes with
same name, same namespace URI, but different prefixes.
Changed auto attribute replacement to depend on local name
and namespace URI and never on qualified name alone

Removed getFirstChild(), getPreviousSibling(), and getNextSibling()
methods from Node

Added indexOf() method to ParentNode

Spell checked the API documentation

Moved XOMResult into the nu.xom.xslt package. 
XSLT still doesn't work, but it's a little closer to working. 


===================
Done 1.0d6:
===============

Now require all namespace URIs to be absolute URI references

Fixed TextWriter bug that prevented the line separator from being changed

Fixed a bug that allowed the namespace URI of a prefixed element to
be reset to the empty string

Fixed a bug that allowed the prefix of an element to
something that conflicts with one of its attributes
or additional namespace declarations

Element.toXML now generates empty-element tags for empty elements

Fixed a bug that prevented the detach() method from working on leaf nodes

Fixed a bug pointed out by Laurent Bihanic in getNamespaceURI(String prefix)
that failed to return namespace URIs from more than one level up in the hierarchy

Fixed a cosmetic bug in the handling of nbsp in ISO-8859-11 Thai

Added a nu.xom.xincluder package to provide XInclude support
The samples package includes a driver program that uses this
to resolve XIncludes in existing documents.

Added a nu.xom.canonical package to provide Canonical XML output.
The samples package includes a driver program that can canonicalize 
existing documents.

Relative URLs in system identifiers for DTDs are now resolved against the base URI
of the document specified in the builder instead of the current working directory.
Apparently this wasn't being picked up from the InputSource. 
I had to add an EntityResolver to take care of this.
Should this really be necessary?
Is this perhaps a Xerces 2.1 bug? 

===================
Done 1.0d5:
===============

Added getName() equals(), hashCode(), and toString() methods to Attribute.Type,
mostly to make it safe for multi-classloader environments

Added a method to Builder to parse a java.io.File object

Added  methods to Builder that allow the base URI to be specified when
  building from a Reader or an InputStream
  
Added an experimental method to Builder that builds directly from a String 
containing well-formed XML. 

Cleaned up Builder internals to reduce duplicate code.

Fixed a Builder bug that was preventing the default XMLReader from being loaded
in some circumstances

Added support for the following character sets to Serializer:

    ISO-8859-2
    ISO-8859-3
    ISO-8859-4
    ISO-8859-5
    ISO-8859-6
    ISO-8859-7
    ISO-8859-8
    ISO-8859-9
    ISO-8859-10
    ISO-8859-11
    ISO-8859-13
    ISO-8859-14
    ISO-8859-15
    ISO-8859-16

Note that although XOM supports them, not all Java virtual machines do.

Serializer matches character set names case-insensitively as specified in the
XML specification.

Fixed a bug in UnicodeWriter that was preventing reserved characters 
such as & and < from being 
escaped when the encoding was some variant of Unicode. (This is more evidence 
that premature optimization is the root of all evil. I just couldn't resist an obvious optimization
in the UnicodeWriter class, and it came back to bite me in the ass.)

Fixed a bug with unnecessary xmlns="" declarations on root elements by Serializer
and toXML in Element

Fixed incorrect hexadecimal escape sequences generated by TextWriter



===================
Done 1.0d4:
===============


Write unit tests for getAttributeValue

renamed getStringForm toXML()

base URI property in Node 

Give Serializer an option to preserve base URIs by adding xml:base attributes

Added missing write(Docype) method to Serializer to
fix a nasty infinite loop
Fixed bug that did not undeclare default namespace as necessary

Could I make the Attributes class non-public by just adding
getAttribute(int i) method to Element?

Ditto for namespaces?

changed addAdditionalnamespace to declareNamespace

Changed readAttribute to getAttributeValue

Moved removeChildren from ParentNode into Element because it didn't work 
on Document.

Added protected methods to allow the monitoring of insertions and deletions
from subclasses of Element and Document

Added protected methods to allow the monitoring of 
additional namespace declarations from subclasses in Element

Added protected methods to allow the monitoring of 
changes of local name, prefix, and URI from subclasses in Element