W3C home > Mailing lists > Public > www-dom@w3.org > January to March 1999

RE: DOM & CORBA

From: Stephen McConnell <mcconnell@osm.net>
Date: Wed, 20 Jan 1999 07:02:01 +0100
To: "Ingargiola, Tito" <ti64877@imcnam.sbi.com>
Cc: <www-dom@w3.org>
Message-Id: <000701be443a$65c84c10$c212c6d4@cybercable.fr>

Ingargiola, Tito wrote:
> > I imagine I'm missing something a bit obvious...  I'm confused
> > by the differences between the IDL definitions
> > (http://www.w3.org/TR/REC-DOM-Level-1/idl-definitions.html)
> > and the Java Language Binding
> > (http://www.w3.org/TR/REC-DOM-Level-1/java-language-binding.html)
> > provided along with the DOM spec.
> >
> > It seems to me that running any corba2.2-compliant idl-to-java
> > compiler ought to generate the latter from the former (as well
> > as a slew of stubs, skeletons and helper files).  This doesn't
> > seem possible, though, as the IDL definitions define different
> > attributes/methods than the Java language
> > binding.  Example:
[snip]

The problem your observing is due to the fact that the W3C have chosen not
to support the OMG language mappings.  I heard several reasons for this
(political and technical) but the only reason that seems to make sense is
that there are certain attributes which need to be expressed as attributes
in order for ECMA script applications to properly represent the DOM object
model.  The problem arises in relationship to exceptions associated with
ECMA attributes which cannot be declared in IDL nor mapped under OMG
standard language mappings.  While I think this is a major glitch in the W3C
spec, there are solutions.

The archive http://www.w3.org/DOM/updates/REC-DOM-Level-1-idl-19990107.zip
contains an improved definition of the DOM in IDL. Look for the file
contained in the zip archive named dom.idl.  This file contains IDL that is
based on the OMG style guide includes correct ifndef and endif declarations
and pragma prefix declarations.  It also aggregates all of the IDL into a
single include file (as per OMG recommendations).

There are still comments in the IDL that state that some attributes generate
exceptions.  A solution to this is on its way thanks to the Component
Submission under the OMG - a submission which extends IDL to include
exception declarations on attributes.  The Component submission will
probably go to the Technical Committee for vote in March this year. You can
take a look at the Component spec to get an preview of the IDL extension and
mappings by downloading http://www.omg.org/docs/orbos/98-12-02.pdf.

> > Is anybody using DOM objects in a CORBA environment?  How are
> > you dealing with this "mis-match"?  What am I missing here?

OSM is working on an implementation of a set of OMG specifications dealing
with Users, Task, shared Workspaces, Resource objects and a set of
facilities dealing with commercial negotiation for electronic commerce.  We
have including the revised DOM IDL in our submission because we want a very
close mapping with DOM over a CORBA environment.  Aside from the IDL
miss-match problem, there is also the problem that DOM is defined in a way
that assumes the implementation is local. We are dealing with both problems
by defining a DomServant, DomAdapter and DomNode that incorporate node level
declarations of access control (based on existing OMG standards), object
identity, transactional behaviour, and a subscribable structured event model
suitable for large scale distribution.  Our spec will be finalised on the 1
March 1999. As soon as it final we'll be posting it to our web site under
http://www.osm.net/infrastructure.html#negotiation.

Cheers, Steve.

Stephen J. McConnell, OSM sarl
Business Components for Electronic Commerce
http://www.osm.net
mailto:mcconnell@osm.net


> > Many thanks for any help!  Regards,
> >
> > 	Tito.
Received on Wednesday, 20 January 1999 01:02:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 22 June 2012 06:13:46 GMT