- From: Michael Kay <mike@saxonica.com>
- Date: Thu, 8 Oct 2009 22:49:06 +0100
- To: <noah_mendelsohn@us.ibm.com>, "'Henry S. Thompson'" <ht@inf.ed.ac.uk>
- Cc: "'C. M. Sperberg-McQueen'" <cmsmcq@blackmesatech.com>, <www-xml-schema-comments@w3.org>
> Henry Thompson wrote: > > > were always intended simply to make it easy for processors > to document > > what they _expose_, not to let them off the hook as regards > > implementation. > Noah replied: > > As to what you implement internally, my answer would be along > the lines of: wrong question. It is often the case that the only way > to correctly compute what you're exposing is to build up > information that is isomorphic to what's in the components, > but there may sometimes be other ways to do it Indeed. What goes on inside the processor is none of our business. The question whether the processor conforms to the specification can only be decided based on the externally visible behaviour of the processor, and not on its internals. It must be possible for a third party to test a processor to determine whether it is conformant, without any access to details of the implementation; if a conformance claim cannot be verified independently, then it is meaningless. So I find the phrase "let them off the hook as regards implementation" rather strange. There is no such hook. Michael Kay
Received on Thursday, 8 October 2009 21:49:54 UTC