Sun evangelist denounces WS-I's ‘hidden agenda’
Never one to seek refuge in the safety of entrenched corporate positions, Phipps laid it on the line. He said he believes that WS-I is about agreeing standards that are proprietary, even if the fact is not explicitly mentioned at the time, in the expectation of asserting ownership - or even charging royalties - at a later date.
The accusation gains plausibility in the light of IBM's recent announcement that it retains patent rights over technology that it submitted to the Organization for the Advancement of Structured Information Standards (Oasis) two years ago. Other members were shocked, and not entirely satisfied with IBM's claim that it had failed to mention the patent at the time of submission due to an oversight.
Dismissing the most popular explanation for Sun's apparent exclusion from WS-I's decision-making board - that IBM and Microsoft just took the opportunity to avenge themselves for past indignities - Phipps pointed out that "Sun is too small to be worth bullying".
But how about the plausible claim that Sun was not invited to be a board member because it simply is not a leading player in web services? That is not so easily waved aside. After all, it was Microsoft (together with DevelopMentor and UserLand) that first submitted Simple Object Access Protocol (SOAP) to the World Wide Web Consortium (W3C) in 1999 and got the ball rolling. IBM, too, was quickly off the mark and joined Microsoft and others in submitting SOAP 1.1 in summer 2000. Meanwhile Sun was apparently absorbed in building the complex edifice of specifications that go to make up Java 2 Standard, Enterprise and Micro Editions (J2SE, J2EE and J2ME).
After initial opposition to SOAP, however, Sun did a quick U-turn in May 2000 when the Electronic Business using XML (ebXML) Group decided to incorporate SOAP 1.1 in its eponymous standard. At the time Anne Thomas Manes, Sun's director of business strategy, said: "At this point we are no longer opposed to SOAP."
There was a lot of interest in web services at the recent JavaOne conference, added Phipps - although it was the potential of mobile Java that really gave delegates the feeling of "a new era dawning". He insisted that J2EE has the most mature interfaces for adding web services, and adds that many JavaOne visitors were incredulous to find that it is "dead easy" to do so. "You just add a little code to your Java classes, and it's done," he said. The unspoken implication is, "So why are all these other vendors making such a song and dance about supporting web services?"
As for WS-I's ostensible mission - to promote web services interoperability across platforms, operating systems, and programming languages - that is very much what Oasis was set up to do, said Phipps. The Oasis "frequently asked questions" (FAQ) document states that "The work of Oasis complements that of standards bodies, focusing on making these standards easy to adopt, and the products practical to use in real-world, open system applications."
Moreover, the standards with which Oasis is primarily concerned are Standard Generalized Markup Language (SGML), eXtensible Markup Language (XML) and Hypertext Markup Language (HTML). As XML is more or less the single fixed point in the highly fluid world of web services, it does seem that WS-I's charteroverlaps that of Oasis to say the very least. So if it was not really needed to coordinate the various web services specifications, Phipps wants to know, why was it really set up?
Is Sun well placed to criticize IBM and Microsoft, in view of its own refusal to hand over control of the Java specifications to a standards body such as the International Standards Organization (ISO) or ECMA? Not fazed in the least, Phipps maintained that Sun's licensing of the Java specifications is used only to enforce compliance testing.
This is a valid practical point, as the Object Management Group (OMG) has found the hard way. Because OMG specifications, such as those for the Common Object Request Broker Architecture (Corba) and the new Model Driven Architecture (MDA), can be downloaded without registration, nobody even knows how many organizations have obtained them. And because OMG does not license its specifications, it cannot require that implementers perform compliance testing. Although The Open Group has a Corba branding process, it is expensive and so far few vendors have chosen to submit their products for testing. On the other hand Sun can easily tell exactly how many products on the market are certified to comply with J2EE 1.3, for instance.
As Phipps explains it, the Java Community Process (JCP) has now been refined to a point where Sun has little or no control over what it does. In fact, he says that Sun's own product groups have no more advance knowledge of how specifications will turn out than any other vendor - obviously a healthy state of affairs for other Java vendors whose products compete with the Sun ONE lineup.
Returning to the subject of standards bodies and their transparency - or lack of it - Phipps singles out the Internet Engineering Task Force (IETF) as "a good template" because it is "very code-centric" and requires that there be at least two working versions of a proposed innovation before it can be considered. In this context, it is interesting to note that unlike IETF, JCP, Oasis and OMG, WS-I's web page contains absolutely no public information about its organization, constitution, process or board members. Perhaps they will get around to it ater.