Re: [JXPath] unordered node-sets

[Knut Wannheden <kn...@gmail.com>]

Dmitri,

Knut Wannheden <knut.wannheden <at> gmail.com> writes:

> 
> Dmitri Plotnikov <dmitri <at> apache.org> writes:
> 
> > @name is a reserved attribute name for Java Objects.  It in fact checks the 
> > name of the property/key rather than the value of the property called 
> > "name". I think it is equivalent to "[name() = 'Foo'].  This feature should 
> > be deprecated, and I will try to do so in a future release.
> > 
> 
> Yes, I think name() is indeed equivalent and IMO preferable as it's more XPath
> style and doesn't keep users from using @name to access an object's property
> "name" (i.e. getName()).
> 
> In the JXPath based EMF search tool I've been working on
> (http://emfsearch.sourceforge.net/) I have now customized the JXPath axes as
> I've previously outlined (child:: and thus also descendant:: only for EMF
> containment references).  With this customization it is now very difficult to
> write queries which access the getName() method of an object.  Instead of
> "[@name = 'Foo']" I have to use something like "[string(@name) = 'Foo']".
> 

If found a reasonable workaround for this which might be useful to others.  The
equality check can simply be turned around.  I.e. "['Foo' = @name]" works fine.

Now I've just got two more questions on this :-)  Have you been able to check if
removing DescendantContext#isChildOrderingRequired() breaks anything?  Do you
want me to open a feature request for either of these two changes?

Regards,

--knut


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-user-help@jakarta.apache.org

Re: [JXPath] unordered node-sets

[Dmitri Plotnikov <dm...@apache.org>]

Knut,
 
Thank you for the feedback.  There is no need to make an "official" feature request. I'll take care of it.
 
Regards,
 
- Dmitri

Knut Wannheden <kn...@gmail.com> wrote:
Dmitri,

Knut Wannheden gmail.com> writes:

> 
> Dmitri Plotnikov apache.org> writes:
> 
> > @name is a reserved attribute name for Java Objects. It in fact checks the 
> > name of the property/key rather than the value of the property called 
> > "name". I think it is equivalent to "[name() = 'Foo']. This feature should 
> > be deprecated, and I will try to do so in a future release.
> > 
> 
> Yes, I think name() is indeed equivalent and IMO preferable as it's more XPath
> style and doesn't keep users from using @name to access an object's property
> "name" (i.e. getName()).
> 
> In the JXPath based EMF search tool I've been working on
> (http://emfsearch.sourceforge.net/) I have now customized the JXPath axes as
> I've previously outlined (child:: and thus also descendant:: only for EMF
> containment references). With this customization it is now very difficult to
> write queries which access the getName() method of an object. Instead of
> "[@name = 'Foo']" I have to use something like "[string(@name) = 'Foo']".
> 

If found a reasonable workaround for this which might be useful to others. The
equality check can simply be turned around. I.e. "['Foo' = @name]" works fine.

Now I've just got two more questions on this :-) Have you been able to check if
removing DescendantContext#isChildOrderingRequired() breaks anything? Do you
want me to open a feature request for either of these two changes?

Regards,

--knut


---------------------------------------------------------------------
To unsubscribe, e-mail: commons-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: commons-user-help@jakarta.apache.org