Why not this:

the XML request PDU is a MIME message with a content-type: text/xml sent via
HTTP: to a particular web page i.e.
HTTP/1.0 POST /page
Content-Type: text/xml

<?xml version="1.0" ?>

and the response PDU is another XML MIME message.

This is fairly simple and gets the job done. The contents of the XML MIME
message body are left to the "page" to interpret.

Jonathan Borden
JABR Technology Corp.

> -----Original Message-----
> From: General discussion of Extensible Markup Language
> [mailto:[log in to unmask]]On Behalf Of Xml Starter
> Sent: Friday, December 04, 1998 7:05 AM
> To: [log in to unmask]
> Subject: Proposal: XML to XML protocol over HTTP
> Ok everyone, here's a vision I want to share (I'm sure I'm not the
> first with it), and want a proposed solution for (I'll propose one
> too).
> Imagine a world in which all the business processes in the world
> happen automatically without human intervention (except of course
> for some things human should do, like some decision-taking).
> In this world, these business processes communicate with each other
> all over the world using XML. However, in order to do this, there
> has to be some for of transfer-protocol-level standard that they
> all agree with so that connecting and doing business over the Internet
> with any company is as simple and easy as connecting today with a
> browser with a website.
> Today for example, a web browser knows that it connects to a port 80,
> and that the protocol spoken there is HTTP. This way ANY browser
> can connect to ANY server, so long as they both understand HTTP
> Today however, it seems we cannot do this with XML, as they way to
> pass XML around is not standardized. i.e.: some people communicate
> between two systems using RMI or CORBA by passing and XML stream
> encoded in an RMI or CORBA string object. This however defies the
> whole purpose of simplicity and it implies that the other side also
> understand RMI or CORBA (or other methods like custom-socket code,
> HTTP POST, etc).
> So what do we need, a standard. Maybe we need 2 standards, an unreliable
> one based on HTTP, and a reliable one maybe specifically created for
> this purpose (like HTTP was created). This however is not a big issue,
> the big issue is that we pick one. I suggest a XML over HTTP as a
> first step because it will probably be the simplest one to implement,
> and we can use the existing infrastructure where things like Java
> Servlets pay a big role and are already widespread and ideal for this
> kind of situation.
> Once we do this we will be able to do things like in the following
> scenario: Company BigSeller says they want to do e-business with
> Company BigBuyer. Well, BigBuyer says, just send me the XML tags
> you're using (including the DTD?), and your Internet address, and
> a description of the flow of information we'll be exchanging and
> I'll take care of the rest. Now a programmer in BigBuyer knows what
> "business-level protocol" BigSeller has defined, and also knows that
> all this stuff happens over port 8080. He/She gets to work and never
> has to ask BigSeller if they use CORBA, RMI, DCOM, whatever, all he/she
> cares about is to know on which port to talk to. So we could say that
> he/she is going to XML://, and the XML on both sides
> is generated dynamically at run time, to represent things like price
> lists, quantity, etc.
> He/She writes the program, and voila it works. The same thing can do
> all the other BigSeller who want to do business with BigBuyer. As easy
> as HTTP and HTML today, but doing much more complex things. Obviouly
> they both might be using strong SSL to encript data, digital certificates
> and so on, but those are minor details to all this.
> So this is my vision. Comments? Is this being done already and I'm
> just ignorant on the latest developments? If this is not being done,
> where can I go to suggest a standard for doing this?
> Thanks
> Saile.
> ____________________________________________________________________
> Get free e-mail and a permanent address at