Dave's History of SOAP

Sat, Sep 25, 1999; by Dave Winer.

SOAP worked with Dave as they knew XML RPC was the existing protocol. It annoys me that everyone wants to be a superstar and invent new protocols without consulting poeple who are already doing it.

That's not exactly true. Before folklore becomes reality, XML-RPC was originally, privately, called SOAP, when Don Box and I were working with Bob Atkinson and Mohsen Al-Ghosein at Microsoft, in early 1998.

UserLand had a protocol before that called "RPC", I announced it in DaveNet, and they asked if I'd like to work with them on this.

I put a hold on our work and posted a heads-up to Frontier developers that the spec might be changing, based on the first meeting we had with the Microsoft folks.

We quickly implemented a client and server for what was then called SOAP, and Mohsen wrote a client and server too, in JavaScript I believe, and we got them working together. Then a lot of other MS people got in the loop, and the arguing began, and it dragged on for weeks.

I talked privately with some of my friends, "What do I do?". We analyzed the choices. I could stay with the program with MS, and now we know where that would have led. We would have waited over a year, with our users in limbo.

We could go back to the "RPC" format, but we had already implemented the "SOAP" server/client, and it was much better than the old one (the old one didn't have <struct>s or <array>s).

Or we could change the name to something else and release it publicly. I asked the MS guys how they felt about this, and they said nothing. So I sucked in my breath, released the spec as XML-RPC, and waited for them to squash me (I was trained by Apple, who definitely would have crossed me off their list for not being their slave). They never squashed, and in fact, they kept inviting me to meetings to discuss this or that about their spec, which was evolving into something that would be hard to see as originating from the XML-RPC spec.

To me, it was most important to get Microsoft out publicly promoting the idea of low-tech wire protocols based on the standards of the Internet. I would have been just as happy with support from IBM, Oracle, Sun, Apple, Netscape, whatever, because I know that the value of a big name is essential in making something like this stick. Once we had a basic agreement with MS on what became XML-RPC I went on a private tour of execs in the industry, but none of them (I think) had a clue what I was talking about. Microsoft, on the other hand, as a group, got it immediately, all the way to the top, I emailed with Gates on this several times.

Internally, they acted like a standards body, with people from IETF and W3C arguing over all kinds of things. I had no time for the arguments. I basically said yes to everyone. Go for it. Let's do it, I kept saying. But it kept dragging on.

Now we're implementing our SOAP stuff, and we'll plug it in behind all the interfaces we're doing. When I say XML-RPC now, I mean SOAP *and* XML-RPC. Any interface I define now will also be a SOAP interface when we get our server running. Bierman is working on that, for now, but I expect that Andre and I will take it over soon.

Anyway, I'm rambling. The bottom line is that people *will* reinvent the wheel. It always works that way. Instead of hating them for doing it, love them for it and make sure that we can hide their differences behind scripting APIs. Then everyone can win, and no one will be threatened by repeating innovation loops.