W3C home > Mailing lists > Public > www-qa@w3.org > August 2002

More about Type of Spec and Class of Product (Spec Guideline 2)

From: David Marston/Cambridge/IBM <david_marston@us.ibm.com>
Date: Tue, 13 Aug 2002 10:36:45 -0400
To: www-qa@w3.org
Message-ID: <OF85495761.D2294782-ON85256C14.004DBA1B@lotus.com>





The latest taxonomy of the types of W3C specs is:
1. foundation or abstract (e.g., Infoset),
2. content/data (e.g., MathML, SVG),
3. protocols (e.g., SOAP),
4. processors (e.g., XSLT, XML Query),
5. APIs (e.g., DOM),
6. notation/syntax (e.g., XPath),
7. set of events (e.g., one part of XForms),
8. rules for deriving profiles (e.g., part of SVG).

Yesterday, Dominique wrote:
>the 4th bullet "processors" doesn't answer the same question as the
>other bullets: XSLT doesn't specify a "processor", but a processing
>language. I would say too that the 2nd bullet would be clearer with
>"content/data formatting"

As someone who works with the XSLT spec as my primary occupation, I
find it pretty clear that the spec intends to define the behavior of
a processor. I think the Conformance chapter of the XSLT 1.0 spec
(Chapter 17) makes this clear as well. Furthermore, I asked Scott
Boag, a member of the XSL WG, if they intended the spec to define the
behavior of a tool that creates stylesheets (generator or editor) and
he said no, just the processor. Thus, a stylesheet generator is under
no obligation to produce an interoperable stylesheet (one that has no
dependencies on processor-specific features or particular extension
libraries).

More generally, Guideline 2 encourages WGs to distinguish between
what will have to conform (class of product) and what the WG has
created. For XSLT, the WG created a set of declarations and
instructions, but the class of product that has to conform is a
transformation processor. Still, it may be worthwhile to adjust the
verbiage of the 8 numbered items by looking at the whole matrix of
specs, which I haven't yet done in depth. The goal is to clarify
that some specs are about "static" content (like MathML), some are
about recipes for action (like XSLT), and some are about dynamic
interchange. The testing strategies will be different, and thus the
test suites will be different.

I think we should use this list to discuss the classification scheme,
but only get really active on it after the next Working Draft is out
for comments.
.................David Marston
Received on Tuesday, 13 August 2002 10:37:38 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Sunday, 6 December 2009 12:13:59 GMT