[Home] [By Thread] [By Date] [Recent Entries]

  • To: xml-dev@l...
  • Subject: XPattern Specification
  • From: Alan Gutierrez <alan-xml-dev@e...>
  • Date: Sun, 16 Jan 2005 17:02:39 -0500
  • Mail-followup-to: xml-dev@l...
  • User-agent: Mutt/1.4.1i

Quoting Uche from an off list message:

> I agree that we should move to XML-DEV and start with a clear
> statement of goals for the XPattern spec.  Seems to me that 6
> foundational goals are:

> 1) Simple
> 2) Subset of XPath
> 3) Cross platform/language/implementation
> 4) Practical for most streaming implementations
> 5) Supports declarative as well as procedural semantics  (the Michael
>     Rhys goal :-) )
> 6) Extensible

    I'd like to discuss point six (6), maybe forking a new thread,
    since I'd like to develop a XPath language binding API. I'd like
    to bind the Java extensions I've written for Jaxon to Saxon.
    Like SAX, the API would be expressed somewhat differently in
    different languages, but would allow reuse of extensions across
    XPath or XPattern implementations.

    That's one thought.

    The second is that I'd like to develop a pattern matching
    engine, so I'd like to see these experssions compile to pattern
    matching automations. I'm not sure how what such a requirement
    will impose upon the langauge. For this application the subset
    of XPath might be larger. I'm not sure it's a good idea, but
    pattern matching seems like a natural for a stream of tokens.

    That's a request.

    I'll also note that more thought put into streaming will mean
    that documents will be designed for streaming applications. In
    discussion, the argument that real world documents don't always
    convientently have the particpants of a predicate preceeding the
    current node in document order is specious. In the real world
    information isn't always in rows and columns, either.
    
    I'm sharing a realization here, FWIW.

    Also off-list, Petr Cimprich suggests:

>   I think we should try to define an EBNF grammar that could be
>   implemented in different languages. Some time ago, Oliver Becker
>   derived an STXPath grammar from XPath2. Here is the diff:

>   http://www.informatik.hu-berlin.de/~obecker/stxpath/stxpath.html                
>    This approach seems to work, we can start again, take XPath2
>    and discuss necessary constrains.

    Here's XPath 2.0:

    http://www.w3.org/TR/xpath20/

--
Alan Gutierrez - alan@e...

Site Map | Privacy Policy | Terms of Use | Trademarks
Free Stylus Studio XML Training:
W3C Member