[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: XQL project update
This all sounds not that bad ;) Thanks Ann, for your work and this status
message. The first ozone based DOM implementation is included in last release
package. (so there is something to play with) It is very slow but so far it
plays with OpenXML. Another problem is the "node explosion". DOM nodes are
often very small and so the db overhead is huge currently. We will work on both
problems. As a next step we consider integrating the DOM stuff into the ozone
core. The goal is to represent the entire db content as DOM. This seems to be a
great way to put an end at the "query language (and ad hoc tools) in ozone"
debate because it will be possible to use any DOM based tool to visualize,
query and manipulate an ozone db. Ann, I think this was your intention from the
start. Sorry, it tooks me some time to get this point ;) _And_ it will be
tricky to implement this. Anyway, together with the XPath we could have a
great foundation to build a complete XML repository.
Falko
On Thu, 14 Oct 1999, Ann Tecklenburg wrote:
> I have found an XPath grammar for JavaCC that
> was generously made available by Ingo Macherius
> (et.al.). Ingo is a principal in both XPath and XQL
> and is a recognized expert in the general field
> of database systems and query languages.
>
> I have looked at the (small amount) of available
> open source code (mostly coming from XSL
> implementations) and decided, for now, not
> to try to directly adapt it for use in the Ozone project
> based on the following reasoning:
>
> The purpose of a doing an XPath implementation
> is to provide ground work for an XML-query
> implementation when the working group
> gets done with their spec. It appears
> that the "XQL" will be based on XPath
> with additional selection and condition
> operations.
>
> If you use a grammar and a CC, you can
> quickly regenerate the parser as the
> specification evolves---and I expect
> that it will do alot of changing before it
> becomes a W3C "recommendation".
>
> BTW, the parser I am referring to is
> an XPath "expression" parser, not
> an "XML" parser. These are
> very different things although XPath
> expressions are elements of XML.
>
> About the only thing that an XPath parser
> and an XML parser would share in common
> is the namespace specification.
>
> I am going to try to use SableCC from
> MCGill University based on the following
> reasoning:
>
> Sable is GPL'ed and has a very nice OO
> structure. It automatically builds a parse
> tree which is an absolute requirement for
> any XML work.
>
> SableCC's grammar is very similar to JavaCC
> and its programming style is very consistent
> with both GNU regexp and Java.
>
> Sable has recently had a number of bug fixes
> applied and has been updated to JDK1.2, so
> it appears to be in a good state of repair and being
> regularly maintained. (I wasn't sure of
> this for awhile.)
>
> I expect to begin work on converting the JavaCC
> XPath grammar into a SableCC grammar about
> Wednesday of next week. I hope to have a running
> XPath parser in less than a week of work---if
> not, it indicates that I have to choose a different
> approach (the Sable attempt will have failed.),
> probably indicating the use of JavaCC.
>
> I plan to donate the XPath parser code to the OpenXQL
> project as well.
>
> (JavaCC is "freely available" but does not and never will
> be "GPL'ed" or have source-code available.)
>
> Once the XPath parser is "parsing", the next step is to
> to build an expression evaluator for the XPath parse
> tree. (Uh! that may be the hard part!)
>
> Comments greatly appreciated.
>
> Best Wishes,
>
> Ann Tecklenburg
> email: at@ingenuity-sw.com
--
______________________________________________________________________
Falko Braeutigam mailto:falko@softwarebuero.de
softwarebuero m&b (SMB) http://www.softwarebuero.de