Low-level EDXL

["Kon Wilms" <kon@datacast.biz>]

All,

I raised this with the few folks that were on last week's call, and they
suggested I put it to the list:

I would like to propose that the EDXL spec be extended to include a
low-level version of the EDXL wrapper implemented as a protocol header
extension.

Specifically this would involve specifying a header extension to the
underlying TCP or UDP protocols. This is not a foreign idea - RTSP and
similar protocols 'bolt on' to TCP/IP and UDP/IP in the same manner.

This would enable EDXL to function as a true protocol-level routing
mechanism, and not only an application-level routing scheme, enabling
systems to route the data by socket-level operations instead of having
to fully decode XML and other data.

The downside is that it will need to involve the creating of one or more
test tools to generate the wrapper and transmit data. I'm happy to take
up this task (I've started doing it already, actually). I believe these
tools should be open source, but I am unsure of how OASIS membership
would impact this.

One other point that needs to be raised is that of mapping the
verboseness of EDXL to this layer. Due to the rate at which things are
changing, I believe that this protocol level version can be approached
as either a tuple-based extension or a fixed-field extension with a few
bytes room for future growth. 

In summary this proposal is to add a low-level packet-header-based
implementation of EDXL as opposed to only the application layer as it
currently exists.

Regards
Kon