savas parastatidis

This is why people have stopped believing in standards

The OASIS WSDM TC has decided to submit the WSDM specifications as an OASIS standard despite the negative vote. This is really bad for Web Services standards. Greg Pavlik blogged about what has been happening in the TC (post 1, post 2, post 3).

I totally agree with Greg. How can we possible have a standard when its building blocks are moving sand? WS-Addressing is not complete and the work on WS-RF is still on going. As Greg pointed out, WSDM spec even uses versions of other specs that don’t exist anymore. This is a big big joke and it seems to me that the only reason this is happening is because companies/groups want to get their work stamped as a standard so they can market their services/products as being compliant. This is so bad for OASIS.

5 responses to “This is why people have stopped believing in standards”

  1. Ron Kleinman
    Spot on. A standard’s document is only as strong as its weakest normative reference.
  2. Mark Little
  3. Mark Potts
    So I cant let you get away with promotion of rhetoric, so had to respond. Lets get the facts straight and correct some misinterpretations. First, Gregs comments with respect to “WSDM is based on disappearing specs” is incorrect (unlike Dan Rather, I like to thoroughly check my facts before I make public statements). The specs he refers to are not going to disappear. There are active links to various specifications in the reference section of the WSDM specification. The fact that WSRF might not point to these specific versions from the TC’s home page shouldnt be misinterpreted or imply that they are going to disappear. Second, WSDM uses milestone drafts of specs that have not been ratified through the standards process is correct. If people think that this is heresy, then, yes, they should protest. I am not saying that this is not a reasonable view of the world. It’s just that its not the only reasonable view of the world. If OASIS and its members wanted such a rule it would have it in its bylaws or at least in its process description by now (which is otherwise pretty thorough), the reason it is not, I believe, is to promote faster definition and delivery of specifications. One of the real criticisms of the development of standards by the industry and customers is that their demand for standardization is met with a glacially slow response. Its simply unacceptable to further serialize the process and not provide any flexibility with regard to references to other work that is clearly in progress. We have to realize that we live in a world there are always going to be differing versions of specifications and we have to adopt where appropriate if we want to make progress. As one person recently said to me if you sit on the sideline waiting, for the perfect stack with the perfect layering and all the dependencies right, to come along you’d better invest in a good cushion to sit on”. Now it appears that there are one of two paths we, as an industry can take, when attempting to define a standard; 1) all the dependencies have to be approved or 2) the proposed spec needs to be implementable in an interoperable way and provide value to the intended audience. The TC picked the second based on priority and customer demand while maintaining 1 where they appropriate. WSDM is not based on “moving specs” but on specific versions of specs for which further version exist or will exist. The TC has expressed its commitment to making use of standard versions of the specification it depends on in later versions of the WSDM specifications. As a side note, this is not the first time an OASIS standard references a draft spec. I looked through existing OASIS standards and the XML Common Biometric Format (XCBF) v1.1 (which became an OASIS standard in August 2003) normatively references a draft version of ANSI X9.96 which only became a standard in 2004. Why? Because it needed this spec and enough had been produced to support their needs that they wished not to introduce unnecessary delay or duplication! To the point that WSDM uses one version of WS-Addressing and its dependencies like WS-Notification use another doesnt look optimal, each message exchange clearly makes use of one version or the other and you can find out which one to use by reading the specification. What is important about this is that the practical aspects of implementation and the appropriate way to implement WSDM are to support the two versions of WS-Addressing for all message exchanges. For all aspects of WS-Addressing used by WSDM, the two versions have the same semantic; it’s just a matter of accepting elements with the same local name but, coming in one of two namespaces. Do that and you won’t have any problem. Its not ideal but again is the best practical solution if you want to deliver something of value sooner rather than later. On the matter of not being able to implement WSDM using the current set of toolkits available, Apache Apollo and Hermes each provide toolkits that work perfectly well with all the versions of WS-Addressing. Toolkits are supposed to help developers and, at this point in time, a toolkit that wants to help developers with WS-Addressing MUST support the non-standard versions because that’s all there is. I am confident that in short order WSDL 2.0, WS-Addressing, WSRF and WSN will all be standards and toolkits will continue to be developed which support these. If people at Oracle or Sun want to wait 2 years to get to interoperable management using Web services that’s their prerogative , but others in the industry (specifically, the major management software vendors: IBM, CA, HP and BMC) as well as customers, want it now. As far as Tim Bray’s comments go, they were amusing to say the least. Im not sure if Tim really doesnt know what his company’s position on WSDM is, but let me help out here. Sun participated in the development of WSDM and supported its direction right up until a multi-billion dollar check from Redmond came in the mail. Now his company is suddenly against WSDM and in support of an alternative proprietary specification (itself based on a bunch of other proprietary specifications) and is supporting a push for a standards organization (DMTF) to rubber-stamp it “as is” and pushing for another standards organization (GGF) to base its work on the result! Sun can play-out the horrified virgin act but, if they really want to take this stance with respect to specifications and the development of industry standards perhaps they might start at home first. Finally, releasing WSDM 1.0 based on the current dependency was not rushed specification. This is what the TC has been openly working towards since WSRF came out over a year ago. There has been plenty of time for Oracle and others to speak up, including committee drafts and a public review period. If you really care about something, you don’t wait for the vote and leave the TC with just a binary choice, you bring things up during the process, especially if you are a TC member, as both Oracle and Sun were! Im not saying WSDM is perfect, there is always room for improvement, but much of what has been posted is far more politically driven rather than being based on true technical concerns or critical feedback. Mark
  4. Mark Potts
  5. greg
    Com’on Mark (Potts): if I was working for HP, I’d be singing the same tune; no point in arguing on that. But you need to re-read my blog post: it’s a reprint from the WSDM list. Are you seriously comparing me to Dan Rather because of the title? That’s a bit silly. Greg