Article:		Kicking out the Cuckoo
Subject:		Does the Web go beyond serving browser clients?
Date:		2002-04-25 14:38:11
From:		Anne Manes
Edd is correct that SOAP has very little to do with HTML, XHTML, WML, or SVG, but then SOAP has very little to do with supporting browser-based clients. SOAP is focused on enabling application-to-application communication over open Internet technologies. Does this constitute a "true business" of the Web? Some folks at W3C obviously thought so 18 months ago when they chartered the XML Protocol working group. But a rising faction now seems to think that it no longer belongs at W3C. They are claiming that Web services don't conform to the Web services architecture. This crux of the argument at the W3C is over the way SOAP uses HTTP. SOAP uses HTTP simply as a transport protocol rather than as an application protocol. But is this type of HTTP abuse really "harmful" to the operation of the Internet? (Most CGI scripts are guilty of the same abuse of HTTP POST application semantics.) SOAP is an XML application. It makes extensive use of XML, Namespaces in XML, and XML Schema. It is often used with XSLT, XPath, XLink, XML Encryption, XML Signature, and other XML technologies. And a number of people have demonstrated using SOAP and RDF together. So there is at least some potential justification to performing Web services standardization work at W3C. WS-I isn't a standards organization. WS-I is only concerned with advancing interoperability between vendor implementations. All of WS-I's work products are dependent on the specifications being developed by the standards communities. I don't particularly care where SOAP and WSDL get standardized, as long as we have a reasonable venue to improve the specifications. Eighteen months ago, W3C wanted to take responsibility for this task, but now there appears to be a fair amount of discontent with that decision. Perhaps it might be more appropriate to move the work to an organization such as OASIS, which specializes in standardizing XML applications. I think it would be better for all involved if the work is performed at an organization that isn't trying to undermine the Web services architecture (which, to be clear, is not the same as the Web architecture as defined by Roy Fielding.)

Previous Message

Next Message

• Does the Web go beyond serving browser clients?
  2002-04-27 20:48:43 Mark Baker [Reply]
  
  "But is this type of HTTP abuse really "harmful" to the operation of the Internet?"

  
  Not to the Internet, but it is harmful to the Web, because POST can't be trusted to mean POST. With POST, the server gets to decide the function that is performed, not the client. That's why it gets through firewalls. Putting a method name in the POST body changes that.

  
  Think of it like drag-and-drop. When you drop a file on the trash folder, you don't also have to specify the method to be invoked.

  
  "(Most CGI scripts are guilty of the same abuse of HTTP POST application semantics.)"

  
  No they're not actually. The vast majority just accept content POSTed to them from a form, which almost always uses POST to mean POST.

  
  BTW, to respond to your subject line, yes the Web goes beyond browsers. It's for any HTTP client. Google isn't a browser is it?

  
  MB