Christian, --- Christian Sell <[EMAIL PROTECTED]> wrote: > Dmitri Plotnikov wrote: > > Of course you are welcome to look at the source code. But I would > > really appreciate it if you told me what it was that you were > looking > > there for. Such feedback would allow me to improve the > documentation. > > Ok. My conclusion so far is that my initial idea of being able to > determine on-the-fly (during parsing) for a given node whether it is > covered by a xpath expression does not work. This is because the > expression may be referencing items that come come later in the parse > > (e.g., child nodes). Therefore I have to have two passes - one to > build > the node tree, and another one to determine the target nodes for the > xpath expressions and treat them accordingly (e.g., set color "red"). > > This reduces my problem to being able to work on a custom object > model. > I know JXPath supports this - but It seems that Jaxen is > significantly > easier to extend - in fact, I had to implement only one class > (DocumentNavigator), and 2 helper classes. This also includes one > extension function. > > also, with Jaxen I was able to easily separate the core functionality > > together with my model support classes without carrying the other > models > around unnecessarily. > > Conclusion: I'll probably stick with Jaxen for now - unless it > completely breaks in the process. BTW, can you name any significant > advantages JXPath has over Jaxen? To me they seem to be almost > identical > in function. There are some differences, but generally speaking you are right - they are similar in function. JXPath supports mixed models, including JavaBeans, Maps and Collections out of the box, Jaxen requires customization for that. Jaxen has a simpler architecture.
If you need to introduce a custom model (as you apparently do), you take a different approach with Jaxen and JXPath, but I think you pretty much have to do the same amount of work. In Jaxen you implement DocumentNavigator and a bunch of iterators, in JXPath you implement NodePointerFactory and NodePointer - it's pretty much a wash. Also, if you can use custom BeanInfo, the JXPath customization is cheaper. Jaxen is a fantastic product, JXPath is not too shabby either if I can say so myself - so it's really a matter of preference. > >>>>BTW, any hints on how to build JXPath from cvs? I am getting lots > >> > >>of > >> > >>>>test failures, and maven.log is full of error messages about > >> > >>invalid > >> > >>>>uses of Jelly tags. > >>> > >>>If you are having problems with the Maven build, try the regular > >> > >>Ant build - > >> > >>>it works like a charm. > >> > >>what makes me feel uncomfortable is the test failures I see all > over > >>the > >>place: > > > > The test failures are caused by a minor issue with Maven where in > > certain version of Maven it does not copy test resources into the > > test_classes tree properly. In order to get the tests run with > Maven I > > typically copy that resource by hand. But like I said, the Ant > build > > including the tests works just fine. > > > > - Dmitri > > hmm, I think I had the same happening with ant.. I have never seen this with Ant. Build "clean,test" and the result should be fine. > thanks, > Christian - Dmitri __________________________________________________ Do you Yahoo!? Yahoo! Tax Center - forms, calculators, tips, more http://taxes.yahoo.com/ --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
