Beta 12 contains a few small bug fixes including the removal of
an unintentional Java 5 dependence, some more API documentation,
a few small API changes (
getIterableAxis now throws
an exception rather than returning null on a bad axis constant;
getOperator has been pulled up into
rather than its subclasses) and two major API changes:
XPath2XMLVisitorclass, and associated methods have been deleted because they were undocumented, untested, and were tightly coupled to the underlying implementation. They may return in the future if there's demand and if someone volunteers to do or pay for the necessary work to bring them up to the standards of the rest of the code base.
matrix-concatextension function has been removed because its license staus was unclear, the originator could not be contacted, and it was undocumented and untested. If someone cares to reimplement it, it could be restored in the future.
The primary impetus for beta 11 was fixing the build process so it once again generates source bundles. A couple of small, almost cosmetic, bugs were also fixed. If you haven't noticed any problems with beta 10, you can safely skip this iteration.
Beta 9 contains some small optimizations, improvements to the documentation, and minor bug fixes. The license should now be the same across all the files.
Beta 8 fixes a couple of bugs in XPath evaluation and optimizes the code in several places. The test suite has been expanded.
Beta 7 fixes a number of important bugs, especially
involving operator associativity, the
functions, and the dom4j navigator. The various root exception classes
and all their subclasses now support Java 1.4 exception chaining even in 1.3 and earlier VMs.
The DOM navigator should compile and run in both Java 1.4 and 1.5 (i.e. DOM level 2 and
DOM Level 3). Namespace handling in DOM is more accurate.
Paths can be begin with parenthesized steps like (//foo)/x.
Beta 7 also features a reorganized, more modular
test suite and expanded and improved API documentation.
Beta 6 makes a few small bug fixes and code clean ups, including
removing an unintended dependence on Java 1.4.
Most importantly it removes a file
IdentityHashMap) that we do not have the right to redistribute.
All prior betas of Jaxen 1.1 should be considered tainted,
and not redistributed in any way. If your project uses
an earlier version, please remove it
and replace it with beta 6. Jaxen 1.0 is not
affected by any of this.
Beta 5 makes a few small bug fixes and code clean ups, especially in the DOM navigator. It also restores some test files that were inadvertently left out of the Beta 4 distribution.
1.1 is a major upgrade that significantly improves Jaxen's conformance to the underlying XPath specs. Even though it's officially a beta, it is a vast improvement over 1.0, and all users are strongly encouraged to upgrade. Witha few small exceptions (e.g. the the document() function has moved to the org.jaxen.function.xslt package and the ElectricXML navigator has been deleted) it is backwards compatible with code written to the 1.0 APIs.
The lang() function is now supported.
All queries return nodes in the correct document order, without exception.
ancestor::* and ancestor-or-self::* axes no longer include the document node
NaN is handled properly.
The mod operator works on floating point numbers.
Navigators are now included for XOM, HTML DOM, and JavaBeans. These are experimental and may not be included in the final release.
Applied patch submitted by Shawn Bayern to fix the booleanValueOf() method.
Added licenses to each source file and a proper manifest to the build at last ;-).
There is now an XPath interface in the org.jaxen package to represent any XPath implementation. So this means that the XPath API of Jaxen is now polymorphic, the same interface can work with any model.
This now means that the org.jaxen.* package represents a purely interface based API to any XPath engine. So it should be possible to implement XPath, FunctionContext, NamespaceContext, VariableContext on any XPath engine if so desired.
The XPath implementation for each model has now got a fully qualified class name. The following code describes how to instantiate an XPath object for each model.
// for DOM XPath xpath = new DOMXPath( "//foo" ); // for dom4j XPath xpath = new Dom4jXPath( "//foo" ); // for JDOM XPath xpath = new JDOMXPath( "//foo" );
The XPath.valueOf() method is now deprecated, XPath.stringValueOf() should be used instead.
Added new extension functions kindly provided by Mark Wilson. They are as follows...
Locales can be specified either using a variable which is a Locale object or using an xml:lang style string which specifies the Locale via a language together with an optional country and variant such as 'fr', 'fr-CA' or 'es-ES-Traditional_WIN'. e.g.
upper-case( @foo, $myLocale ) upper-case( /foo/bar, 'fr' ) lower-case( foo, 'fr-CA' ) upper-case( @foo, 'es-ES-Traditional_WIN' )
The translate() function is now implemented - thanks to Jan for that!
Some auxillary implementation detail changes, which shouldn't affect the public API in any way are as follows
Initial beta development cycle. Please see CVS changelogs for up-to-date list of changes.
lowercase-case()could well follow the example of XSLT 2.0 by using the Unicode case mappings