Clay Shirky wrote:
>
> ...
>
> > Of course it is trivial to generate both and build them into the same
> > program but that has always been true for HTTP also! The code to parse
> > an HTTP request and response is nearly the same. After the first line it
> > is a simple MIME message!
>
> It may be trivial, but no one has done it.
That's not true. Morpheus, Kazaa and other programs on the FastTrack
network use Web protocols in a peer-to-peer fashion. So does KnowNow.
And of course anybody using XML-RPC or SOAP *on top* of HTTP.
> ... I'm not saying that there
> is some deep difference between SOAP clients and Web browsers, I'm
> saying there's a shallow but important difference.
>
> Here's a prediction, based on my earlier claim:
>
> SOAP-aware clients and servers will have richer conversations than Web
> clients and servers did, because programmers will use the ability of
> both sides of the conversation to read and write SOAP in ways that
> they didn't when data went up as an http request, and come back as a
> MIME-type plus html.
Okay, I didn't understand that when you were talking about "the Web" you
really meant "web browsers." The Web infrastructure is exactly as
amenable to peer-to-peer as the web services stack.
Here's the part I have a problem with:
"With Web Services, if you can read SOAP you can write SOAP. If you can
write SOAP you can read SOAP. So modulo only the firewall issue,
everybody's a peer."
The question isn't whether you can read or write the format, but whether
you can *accept connections*. So this is an equally true statement:
"With the Web, if you can read MIME/HTTP, you can write MIME/HTTP. If
you can write MIME/HTTP, you can read MIME/HTTP. So modulo only the
firewall issue, everbody's a peer."
As further evidence of the equality of the statements I offer the
Mozilla xml-rpc implementation:
http://www.mozilla.org/projects/xmlrpc/
Client, no server, like the rest of Mozilla, because Mozilla doesn't
accept connections.
Paul Prescod
