- From: Alex Rousskov <rousskov@measurement-factory.com>
- Date: Tue, 1 Jul 2003 12:25:51 -0600 (MDT)
- To: Dan Connolly <connolly@w3.org>
- cc: Jeremy Carroll <jjc@hpl.hp.com>, www-qa@w3.org
On Tue, 1 Jul 2003, Dan Connolly wrote: [ skipped the parts were I introduced explicit context to explicitly show that implicit "or else" clause can exist, depending on the spec ] > > And, vice versa, the rollercoaster example can be rephrased as > > > > Rollercoaster riders are 4 feet tall > > But if a kid under 4 feet got on the roller coaster, that would just > become false. And if an XML document contains an unquoted value, the "every attribute value in an XML document is quoted" statement also becomes false! It is symmetric. For a given context, every definition (X is Y) can be rephrased as a requirement (X MUST Z) without any objective harm or gain. So your "or else" criteria does not work well (for me). I wish it would! If you argue that an XML document with an unquoted value is not an XML document (by definition), then I can argue that a rollercoaster rider under 4 feet is not a rollercoaster rider (also by definition). > It wouldn't say whose fault it is that it became false. The reason > we use MUST for agents to make it clear whose fault it is when > protocols are violated. I agree. Please note that I am not advocating mandatory use of MUSTs in every spec. Quite the opposite. I am just saying that without a context we cannot clearly define the difference between what should be a definition and what should be a requirement. So, I argue, let's leave the choice up to the spec authors and mandate only higher/meta-level things. > > Overall, it seems to me that the whole issue of "correct" RFC 2119 > > keyword usage is moot. There is no a single correct way and it is > > not important to have one. > > I disagree. There are better ways and worse ways to use RFC2119. I agree. I doubt a single "best" way can be defined for all environments though. And that is what some proposed versions of SpecGL checkpoints may imply or even require, causing the debate. Alex. -- | HTTP performance - Web Polygraph benchmark www.measurement-factory.com | HTTP compliance+ - Co-Advisor test suite | all of the above - PolyBox appliance
Received on Tuesday, 1 July 2003 14:25:55 UTC