From: Ian Schneider <Ian.S<chneider@ar...> - 2003-06-13 17:58:36
The xpath and features discussion at the last IRC had two camps. The
first said xpath should be implemented in Features. The second,
including me, said it should be an external implementation.
The external implementation has several benefits.
1) Keeps feature clear of ugly code.
2) Because Feature does not (and can not) hold references to parents
(either Feature or FeatureCollection) no parent or sibling axes are
possible (with an internal implementation)
3) XPath objects can be compiled and reused.
4) An XPath query could return several different items, including a set
of Features or a set of Attributes.
I have started work on the external implementation. This work was
greatly simplified by a javacc grammar for an xpath abstract syntax
There is a demo here:
Read the whole page... and let me know if there are browser problems. I
tried a jnlp launcher but there is some strange class verification
problem (I'm using reflective proxies) which doesn't allow it to work
From: James Macgill <jmacgill@ps...> - 2003-06-16 15:04:02
At 11:51 AM 6/13/2003 -0600, you wrote:
>The xpath and features discussion at the last IRC had two camps. The
>first said xpath should be implemented in Features. The second,
>including me, said it should be an external implementation.
The demo is indeed very cool. What I would like to see the demo do in
order to be convinced (which I probably am anyway) is:
Return the actual attribute values if I query them
Work with nested features (not that we have any yet...).
The first should be easy, I'm less clear how an external implementation
will manage finding features children without at least a little support
inside features themselves.