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

Re: [xml-server] Re: Mapping RBD->XML. : Merging DOM

Eric van der Vlist wrote:
> Hi Paul,
> >
> > ... What about merging  some of XML files into one ? ;-)
> >
> > I think that'l be  very interesting to share the usecases
> > here. I wish some day we'l get some more from other
> > participators of this list.
> :)
We are developing xmlBlaster, an open source MoM (message oriented
based on CORBA (and soon other protocols) and on XML to XPath-query
the messages (and client meta infos and ACL as well).

As the messages may contain any data format, the message
content is a BLOB (supplied with a mime type to know how to handle it).

Every message unit has xml based meta informations (we call it
XmlKey) to query messages with XPath (or with its OID).

All 'XmlKey' informations are merged into a big XML DOM tree,
over which the XPath query runs.

It works fine, but we haven't yet tested it with very big
amounts of messages.
One idea we are discussing to speed XPath queries,
is to 'normalize' the big XML DOM tree when merging new xml meta data
into it, as in most cases the meta data are from a similar
problem domain.

What we are still missing is a performant persistance layer for our
big DOM tree. Something like IPSI XQL engine:


But this is not in source code and not free for commercial
use, so we can't use it for xmlBlaster which runs under the LGPL

We are thinking of implementing an own persistent layer,
which should be very performant (disk space is cheap)
and support XA transactions, so xmlBlaster
may run controlled by a Transaction Processing Monitor.

We are following the ozone project discussion, if this
can be a persitence solution for us in the future,
but we are not shure yet.

You find more informations about this project here:


Our problems are not quite the 'perfect xml server' problems,
but it may be interesting to compare.

hope this is interesting for you,


Marcel Ruff
ruff at swand.lake.de