[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re:[Fwd: [xml-server] Prototype. 2 hot issues.]



On Tue, 26 Oct 1999, Eric van der Vlist wrote:
> Hi, 
> 
> > Another mail from the xml-server list. Much of the ozoneXML components are
> > mentioned.
> 
> Looks like you've been faster to discover our list than me to discover
> the ozoneXML project ;=)
> 
> As you've seen, one of the 2 key questions at the moment for us is :
> > 2. Is there any free "persistent DOM" implementation ?
> and we would be very interested to learn more about your project.
The main future directions of the ozone development are 
1 - ODMG support (maybe EJB container on top of ozone later)
2 - XML repository

We have started playing with DOM by implementing a persistent DOM based on
ozone (Java OODBMS). This is already part of the current distribution. But it
cannot be used in real world applications yet. We wanted to show that it is
possible to implement a persistent DOM on top of an OODBMS (ozone) and that it
can be done very fast. There are two problems with the current implementation:

1 - it's very slow.
Today each node is a database object. So each method invocation on a node is a
transaction!... Once we have high level tools like Anns XPath implementation
that can run inside the server, the whole thing will become much faster.

2 - node explosion.
Again, each node is one database object. To come around this problem some kind
of clustering of nodes is needed. This functionality must be provided be the
database server, it cannot be put in the DOM implementation. So we have started
to re-work the cluster management of ozone to meet the needs of applications
that deal with very small object. Most typical application fall into this
category anyhow.

> The other key question is about the validation of the language we're
> considering to use.
> 
> If you can have a look at the following thread :
> http://www.egroups.com/group/xml-server/22.html
> and give us your feedback, it would help us a lot.
Another ozone developer should answer this. (Ann?)

> You'll see that it's quite different from XOQL, one of the reasons being
> that we would like to extend XSL-T to allow updates on the DOM itself.
> 
> I wonder if both are not complementary...
> 
> What do you think ?
Ann?


Falko
-- 
______________________________________________________________________
Falko Braeutigam                         mailto:falko@softwarebuero.de
softwarebuero m&b (SMB)                    http://www.softwarebuero.de