- From: Shelby Moore <shelby@coolpage.com>
- Date: Sun, 05 Jan 2003 03:26:14 -0600
- To: www-style@w3.org
I hope that reasonable people in this list could agree on some compromise axioms: 1. Specification _defines_ semantics. Specification also controls _conforming_ semantics, because _conforming_ implementation must create semantics which conform to specification. 2. Specification also _defines_ non-conforming semantics. However, specification does NOT _control_ non-conforming semantics, because non-conforming implementation does not create semantics which conform to specification. By definition, it is out of control. 3. Since no implementation can be perfect (bugs always exist), then no implementation is perfectly conforming. Even our ability to measure conformance will have bugs. Even our specifications will have bugs. Thus all implemention is to some degree out of control. 4. Since no implemention is perfectly conforming, then specification never perfectly controls semantics. 5. When new tags are created (especially when there is only one implementation), then the implementation semantics is the semantics. Whether the implemented semantics are conforming, depends on the condition of any specification. It is possible that there will not be a well developed normative specification for thousands of new tags created by millions of authors using XBL or XSLT. If we can agree on these statements, then I can agree with those who say conformance to normative specifications is very important. And I will also be able to show why putting the semantic implementation of new tags at the markup transformation layer is superior to XBL. Before I proceed to show that to list, I need reasonable agreement on issues above. -Shelby Moore
Received on Sunday, 5 January 2003 04:25:46 UTC