Gianugo Rabellino wrote:
>
> It took me (much) longer than I expected (unfortunately paid work is
> something you can't leave aside in these days), but finally I have a
> working first implementation of a pseudo-protocol based con the XML:DB API.
How cool!
> Usage is as follows:
>
> 1. grab xmldb.jar, put it in the lib/ directory in order to compile the
> sources, build Cocoon as usually;
ok
> 2. edit cocoon.xconf and go to the source-handler section. Add a
> configuration like this one:
> <source-handler>
> <protocol name="xmldb"
> class="org.apache.cocoon.components.source.XMLDBSourceFactory">
> <driver type="dbxml" class="org.dbxml.client.xmldb.DatabaseImpl" />
> <driver type="exist" class="exist.xmldb.DatabaseImpl" />
> </protocol>
> </source-handler>
Ah, you're connecting to XML:DB API then. Great.
> 3. start playing with it as you do with the cocoon:// and resource://
> protocols:) Use a JDBCish URI such as
> xmldb:dbxml://some.host.running.dbxml/a/collection/ or
> xmldb:exist://some.host.running.exist/a/collection/aresource.xml (the
> idea is that if you want to query a collection you should put a trailing
> slash). The idea is that the driver actual implementation (dbxml and
> exist) is looked up in the standard XML:DB URL format and the
> implementing class is taken from the configuration accordingly.
Is there any XPointer query capability for the XML:DB API? I mean, how
hard would be to do something like
xmldb:dbxml://host/collection/resource#//section[@title="Blah"]
Moreover: is there a way to get a "single file" view of an XML database
in order to do something like
xmldb:dbxml://host/#//news[@date > "20011111" and @date < "20011118"
and @approuved]
(which is what we need if we want to make this publishing system rock
this planet!)
> As usual, please consider this as a donation to the ASF and feel free to
> decide about incorporating it in Cocoon.
Count on it!
> You might want to review the
> code, since this is my first implementation of a Source.
>
> Some random thoughts about the XML:DB stuff:
>
> 1. until now the compilation of the XML:DB stuff had been conditional.
> Given the fact that Apache itself will sponsor the Xindice project, I
> was wondering if it could be the case to have it compiled by default:
> all that needs to be done is adding xmldb.jar to the lib directory
> (which is less than 9K in size).
I'd say +1 to this. Our support for the XML:DB API will make both
projects stronger.
> 2. Given that the XML:DB API have no support at all for metadata, it's
> almost impossible to implement a caching algorithm (there is no way to
> tell what is the document size or the last modification date). I was
> thinking about the Avalon resource monitoring stuff, could it help in
> this case?
Hmmm, have to think about it...
> 3. The next thing to do is to add XPath support.
Ah, should have been reading more but I got excited by the title :)
> The code foundation is
> there, but an agreement is needed about the URL format. Given that an
> XPath query can be run only against a collection, I was thinking about
> implementing it as
> xmldb:adriver://ahost/acollection/?xpath=/my/xpath/query. But I'm very
> open to suggestions :)
I'd suggest using the XPointer spec for this (but might require the use
of "ranges" which might not being supported)
> 4. If this pseudo-protocol proves to be useful and bug free the XML:DB
> generators might have little or no meaning: since they haven't made it
> into any RC this can be a proper timing to decide what to do with them.
>
> Enough for tonight, I'm eagerly waiting for your feedback, expect other
> patches in the days to come :)
Thanks, dude. Nice job!
Now that we'll have XIndice inside and a stable and open API to connect
to, we have total control of both sides.
We just need to add metadata capability to XIndice (to allows
versioning, timestamping, workflow management and such). Hopefully the
new project will start RSN.
--
Stefano Mazzocchi One must still have chaos in oneself to be
able to give birth to a dancing star.
<[EMAIL PROTECTED]> Friedrich Nietzsche
--------------------------------------------------------------------
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
