On Fri, May 30, 2003 at 05:38:23PM -0700, Kip Hampton wrote:
> John Merrells wrote:
> 
> >The query language is XPath, which can be used to select matching
> >documents in a repository, or to select nodes from the selected
> >documents.
> 
> Just off the top of my head, I can think of several cases and ways that 
> dbxml and AxKit could be useful together:
> 
> 1) Replacing (or adding to) the XML content stored on the fliesystem. 
> Here, an AxKit Provider would be the mechanism used select data from the 
> dbxml data store for processing and delivery to the client. As Tod said, 
> the the open issue is how a given request URI is mapped to a given set 
> of data. A one-to-one mapping between, let's say, the request's 
> path_info and the internal dbxml document name is easy, but then you 
> lose the ability to combine data from different containers/documents 
> (which is a huge loss, obviously). Is there a way to set up more robust 
> mapping between a given URI and set of XPath statements that would be 
> used to construct the content that is a) generic enough to meet a 
> significant set of common use-cases and b) not so complex in its mapping 
>  definitions that it creates a barrier for casual usrers? I'd like to 
> hear more about peoples' expectations...

I have some very large documents that I preprocess and split up into
smaller documents to keep things light.  It seems with dbxml I wouldn't have
to preprocess and just query the parts of the document and still
keep things light.

> 
> 2) An XSP and/or SAXMachines Taglib.
> The interface for this seems like it would be more obvious. For example, 
> with the taglib installed and registered (or appropriate SAX Filters 
> included in the processing chain) the following:
> 
> <dbxml:query select="/some/xpath/statement"/>
> 
> would be replaced by the markup returned from the query contained in the 
> select attribute. I can see how something like this could be damn handy 
> for a lot of people. Thoughts anyone?

xinclude + xpointer?
<xi:include href="/path/document.xml#xpointer(//section)"/> 

> 
> Another thing to consider: dbxml containers can have arbitrary bits of 
> metadata-- I can see cases where entries in that metadata could be used 
> to define which style(s) AxKit will apply when transforming the content 
> for delivery to the client (a less polluting alternative to in-document 
> xml-stylesheet processing instructions, if nothing else). How else could 
> we abuse^H^H^H^H^H, er, take advantage of this extended metadata to make 
> things cooler for users of both AxKit and dbxml?

Storing the cache in dbxml seems like a natural fit, especially with the
metadata.  Has there been any more talk/work on a new Cache system?
Could this help with the incremental caching ideas?

Someone on #axkit-dahut was talking about a dbxml provider and cache
module a few month ago ... has there been any progress?

Ed.

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to