- From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
- Date: Mon, 3 Dec 2007 14:41:21 +0100
- To: public-html@w3.org
Maciej Stachowiak wrote: > On Dec 2, 2007, at 4:26 AM, Dr. Olaf Hoffmann wrote: > > The best method to realise something is to specify first, > > then implement it in a useful way and use it. > > If it is the other way round this mainly causes confusion, > > wasted time and a lot of superfluous nonsense in the web ;o) > > I think this is the core of where we disagree. > > Often, the most successful W3C specifications have been the ones that > standardize things that are already implemented, or that are informed > by and updated in accordance with implementation feedback. This does > not always lead to total conceptual purity. But it does lead to > pragmatic solutions that actually get used. > > On the other hand, specifications created by committee from whole > cloth in a vacuum tend to be overengineered committee-designed > nonsense, and frequently fail in the marketplace. The waterfall > process for software development is obsolete. > > Regards, > Maciej Both extremes are suboptimal. Such a language is developed not just as a concept and it is something different than just one specific software version. The documents realised in the language persist partly for many years. Having only theory and not applications has the tendency not to discover technical problems or practical inconsistencies between different part of a larger specification. Having only implementations and describing what they currently do, having such an historical approach is no concept for a language useful for a longer period of time. Experts for technical stuff often getting lost in (indeed important) details, but are not able to see the general purpose anymore. Experts for concepts have often no impression about the technical difficulties of implementation and realisation. Both general concept/overiew and technical elegance in details is needed for a specifcation easy to understand, to use and to implement. And therefore the W3C has such a concept for feedbacks to drafts. Ok - this time it started not with a theoretical concept, but with technical details. Therefore hopefully the working group will get much more feedack for the next official working drafts with more weight on the general concept and semantics to get both aspects of the problem covered by one draft ... If this works, maybe HTML5 will persist as long as HTML4, if not, there will be maybe HTML 5.1, 5.,2 5.3, 5.4 etc, one each year with updates for all new requirements disovered each year, but not covered by a specification too close to the implementations and fashions from the last year.
Received on Monday, 3 December 2007 14:01:30 UTC