Это довольно узкая тема, интересная немногим. Гораздо увлекательнее обсуждение (1, 2, 3, 4, 5), из которого я надёргал дальшейших ссылок.
Перезапись XPath.
Типа //p[../div] --> //div/p
XPath: Looking Forward
Dan Olteanu, Holger Meuss, Tim Furche, Fran[E7]ois Bry
Research Report PMS-FB-2002-4, 2002
The Electronic Library journal, volume 20, number 4, 275-287, 2002
Нахождение предков при вычислении XPath
Since SXPath already _violates_ the XPath evaluation context, we can at least make this _violation_ more suitable for the SXML data model. Instead of storing just a "root node" in the SXPath evaluation context, we can store the whole sequence of context node"s ancestors:
(context node's parent, context node's grandparent, ... , root node)
There is than no need in the expensive searching for a parent from the "root node".
Запоминание узлов, используемых в дальнейшем
It's a neat technique, conducive to high-performance XPath evaluations. XPath evaluation is an active research area -- for example, google for Dan Suciu.
http://www.cs.washington.edu/homes/suci
Прототипчик на основе обсуждения
http://user.rol.ru/~satura/sxp-prototyp
zippers
How interesting! Rather than adding parent-pointer annotations to the nodes in the source tree, you carry annotations in an "intermediary tree" -- the node list being passed from one SXPath step function to another.
...
The idea of "context" is quite remarkable. It reminded me of "zippers" that I learned of recently:
http://citeseer.nj.nec.com/hinze01web.h
(see Section 2 of that article).
A zipper is a way to "mutate" a complex data structure declaratively. ...
Deleted comment
August 20 2006, 17:25:59 UTC 5 years ago
По поводу единого синтаксиса -- в DSSSL есть какой-то язык запросов, но я его не знаю.
Anonymous
July 3 2007, 17:23:14 UTC 4 years ago
free ringtones
Pleasant design, successful navigation.