- From: Jonas Sicking <jonas@sicking.cc>
- Date: Thu, 03 May 2007 04:12:35 -0700
- To: Gareth Hay <gazhay@gmail.com>
- Cc: Henri Sivonen <hsivonen@iki.fi>, Roger Johansson <roger@456bereastreet.com>, HTML WG <public-html@w3.org>
Gareth Hay wrote: > > > On 2 May 2007, at 12:49, Henri Sivonen wrote: > >> On May 2, 2007, at 14:42, Henri Sivonen wrote: >> >>> You might have more success with your argument if you showed using >>> Game Theory, why in a multiplayer "game" (i.e. market) none of the >>> "players" (i.e. vendors) have an incentive to be more permissive >>> towards the legacy baseline of permissiveness than the others. >>> >>> In addition, it would help to show an incentive for the first vendor >>> to move from the status quo towards less permissiveness without all >>> the vendors moving in lockstep. >> >> Just to elaborate: >> It seems to me that undefined error handling corresponds to no >> equilibrium, Draconian error handling corresponds to unstable >> equilibrium and defined error recovery corresponds to stable equilibrium. >> > But we live in a real world, sadly. > We can define away at an error recovery procedure, but we all know the > reality is that UAs will all handle errors the way they want to, or is > more convenient to them, and each one will handle this differently, and > we will be no further on. That is not true at all. UAs follow as much of the spec as they can. Error handling is just like any other part of the spec. The reason UAs don't implement parts of the spec is when it's impractical for some reason, such as constrained on resources (i.e. spec takes a lot of engineering time to implement) or implementing the spec simply does not work (spec is wrong, too many pages out there would break, etc) / Jonas
Received on Thursday, 3 May 2007 11:12:41 UTC