Two different stories this week highlight what’s worked and what hasn’t at the web’s biggest standards body. Elliotte Rusty Harold’s RELAX Wins has generated a stir about how much “W3C XML Schemas (XSD) suck,” according to Tim Bray. On a positive note, though, Norm Walsh announced eight Proposed Recommendations around XQuery, XSLT 2, and (hooray!) XPath 2.
In some ways, though, the story of the shift from XSD to RELAX is similar to the story behind XPath 2 (really the core of both XQuery and XSLT 2). An earlier standard achieved some success, had some critical failures, and was revised into a more usable form. XPath took the more normal form and just bumped a major version number, whereas RELAX had to start totally from scratch, but they both can be seen as the W3C working fairly effectively in the long term.
The importance of the W3C in the eyes of vendors is still as high as ever, as shown by this snippet from the maintainer of a popular XML editor last week shows:
| Is there any way to use the new Saxon 8.8 as XMLMind’s XSLT engine?
| For one of our use case we would need to have its XSLT 2.0 and the SQL
| extensions available.
For now, we really do not intend to upgrade to Saxon 8+. Also note that
XSLT 2.0 is not yet a W3C recommendation. See http://www.w3.org/TR/xslt20/
[UPDATE: A more balanced assessment of the place of XSD by Bob DuCharme]
[UPDATE2: Oops, a stupid error on my part: attributing RELAX to W3C instead of ISO. A good Family Tree of Schema Languages has just been updated by Rick Jelliffe.]
Now that's funny.
Relax (in standard form) comes from ISO fella, or really, originally is the combined work of two separate works from people who didn't think the W3C was up to it because, well, the W3C wasn't up to it. And the W3C doesn't write standards. They write specifications. You do know the difference, right?
The W3C defenders are starting to sound sort of defensive and stodgy about this issue, which is exactly what they were saying about the ISO defenders fifteen years ago while they were burgling ISO standards to write W3C specifications. The truth is that even if some vendors still go to ISO for markup application specifications, they do it because the W3C controls the XML specification. When it comes to the application language specs, it's anyone's game who cares to try. There are dozens of them out there. Schemas written in XML (what XSD is) are application languages. There can be as many as there are approaches to the tasks of validation, grammar checking, assertion checking and so on.
C'mon kid, what really sucks is when you wake up one day to discover you've become your parents. And you have!
HA!
len
Len said:
> "And the W3C doesn't write standards. They write specifications."
If you want to be pedantic, the W3C is a standards body that publishes Recommendations.
http://www.w3.org/Consortium/
"W3C primarily pursues its mission through the creation of Web standards and guidelines. Since 1994, W3C has published more than ninety such standards, called W3C Recommendations. "
@Keith,
Uggg! Just because the W3C doesn't know the difference between a specification and a standard, doesn't mean you have to take it at face value.
Definitions,
A specification: A definition of a particular way to achieve a desired end result.
A standard: Something that enough people have adopted as a particular way to achieve a desired end result to suggest that enough critical mass of adoption has been achieved to suggest that this is "a STANDARD way to achieve the desired end result."
There is NO SUCH THING AS A STANDARDS BODY! There are consortiums of individuals and companies who will, at times, work together to create a desired end result, writing these as SPECIFICATIONS with hope that enough people will adopt this specification such that someday it can be considered a STANDARD.
An example,
XSLT and XPATH 1.0 are SPECIFICATIONS recommend by the W3C as as a STANDARD way to reference and transform XML. LOTS and LOTS of people have adopted them. Since they were first RECOMMENDED as a STANDARD way to reference and transform XML in 1999, they have become the defacto STANDARD for achieving such desired results.
XSLT and XPATH 2.0 (and XQUERY for that matter) are SPECIFICATIONS recommend by the W3C as as a STANDARD way to reference and transform XML. LOTS and LOTS of people *HAVE NOT* adopted them, so it is YET TO BE SEEN whether or not they will become the defacto STANDARD of referencing, transforming, and querying XML documents.
The confusion between the two terms usually stems from the notion that a SPECIFICATION can be used as a STANDARD approach to accomplishing a desired end result. Yep, it sure can, especially when it comes from a consortium such as the W3C or ISO or IETF or IEEE or ECMA or OASIS or etc... who has a proven history of developing SPECIFICATIONS that have since been adopted as STANDARDS. But just because these consortiums have proven they can do it once, doesn't mean they can do it each and every time they release a SPECIFICATION.
In short, a SPECIFICATION != a STANDARD, though it holds the potential to become one.
| Just because the W3C doesn't know the difference between a
| specification and a standard, doesn't mean you have to take
| it at face value.
In fact, that was exactly the intent of the original post, to show one spec (XSD) working it's way to becoming a standard way of specifying the structure and content of XML documents and despite that standardization not necessarily being the best way of doing so (for the Document guys rather than the Data guys, perhaps moreso...).
On the issue of there being no such thing as a "standards body", you've got to be kidding, as you've provided a wonderful definition yourself:
| There are consortiums of individuals and companies who will,
| at times, work together to create a desired end result, writing
| these as SPECIFICATIONS with hope that enough people will adopt
| this specification such that someday it can be considered a
| STANDARD.
@Keith,
>> On the issue of there being no such thing as a "standards body", you've got to be kidding, as you've provided a wonderful definition yourself: > "Standards bodies" don't write standards. They write specifications. Whether or not these specifications reach critical mass in regards to their adoption rate such that they can be considered a standard requires time. In and of itself, this is my point.
| None-the-less >> "Standards bodies" don't write standards.
| They write specifications. Whether or not these specifications
| reach critical mass in regards to their adoption rate such that
| they can be considered a standard requires time.
To that I'll wholeheartedly agree. Thanks for getting me to be more explicit.
MDP,
>>"Standards bodies" don't write standards. They write specifications.
So, ISO writes specifications?
http://www.iso.org/iso/en/aboutiso/introduction/index.html#ten
Or, perhaps, ISO develops specifications, with expert knowledge, to the point that they are STANDARDS (uppercase), so that at some point they becomes standards (lowercase) after enough widespread use?
@Mike,
>> Or, perhaps, ISO develops specifications, with expert knowledge, to the point that they are STANDARDS (uppercase), so that at some point they becomes standards (lowercase) after enough widespread use?
Sounds reasonable.
It is about correctness first. You corrected your error. Good.
Second it is about a narrower definition for the term 'standard' so that it can't be used to beggar the question of acceptance. A standard should be written when a technology has reach and scale such that concerns of interoperability (not portable data, but systemic interoperation) is desirable. A specification is an agreement to develop a technology with some desirable set of features that either does not currently exist or does not have reach and scope.
Don't try to apply that using historical examples from the W3C where most of the specifications were claimed to be standards where standards did exist. In these cases, the word for that is 'burglary' and it is a profitable business particularly where culture applauds it. But if you want a saner and less politically charged effort where the early claims don't lead to more disappointment and the declining fortunes of the organization, be strict in the definitions for the legal claims. As Dr. Goldfarb taught us, "Conserve nouns."
But today the W3C's fortunes are on the wane. It has other business consortiums have before it, served its purpose for its dues paying members somewhat as Freemasonry did in the founding of the American government. And just as such gentleman's clubs come and go in popularity, this one is too. It won't disappear, it is replicated by the Eagles, the Moose and the other clubs musicians sometimes refer to as 'the animal gigs'.