Re: SOAP and State (was: Re: No consensus on draft findings on Unsafe Methods (whenToUseGet-7))

David Orchard wrote:
> 
>.... I just don't think that
> not doing GET as a required binding makes web services in a state of sin and
> in dread contradiction of the web, and therefore they should be booted from
> w3c.  For some reason, that just seems a very bizarre and naive suggestion.

I don't see how it is worth the effort to argue about the consequences
for failure to do something that we agree is technically possible and
worth doing. Let's just do it and move on.

>...
> Paul, I don't want to spend too much time on this. 

You're too busy to listen to me tell you why SOAP will fail and how to
save it. "Driving WAY TOO FAST to have time to look at the map."

3 years from now Web services will be built around hyperlinks, just as
today's Java programs are built around references and today's relational
databases around foreign keys. I'll hysterically suggest that SOAP can
either "invent the future" or be run over by it.

> ... Here's yet another
> example of where revisionist history is happening.  Most web sites do NOT
> use URLs for state management, despite all the claims that they can.  

Web sites do this to precisely the degree that they want application
integration. Expedia wants to be able to email me a URL for an itinerary
so they set a GET-table URL. That URL is the "integration" between the
Web and my mail program. Slate wants articles to be Google-indexed and
otherwise linkable so it gives them a URL. That URL is the integration
between Google and Slate.

> .... Most web sites use cookies for this purpose.  

Many of today's Web sites use cookies for state management because many
sites do not need to share state. They are also impossible to integrate
with other sites. That's what web services is supposed to fix.

But if you want to replicate the flaws of today's web go right ahead. ;)

> ... If we made "SOAP Cookies", or "SOAP
> Biscuits", that functioned similarly, would that solve your remoting issue?

No. I think you know that cookies are completely different than, and
less powerful than, references. What would a database person say if you
offered them cookies instead of foreign keys.

I'm going to skip all of the procedural and historical stuff because the
main point is the technology.

 Paul Prescod

Received on Friday, 26 April 2002 06:30:56 UTC