On 8/18/2011 9:05 AM, Anne van Kesteren wrote:
On Thu, 18 Aug 2011 15:00:28 +0200, Jeff Jaffe <[email protected]> wrote:
I would add to this discussion that we need to be mindful of where we
are in the development cycle.
Certainly, in the early phases of development it does not yet make
sense to converge to a fixed point. We have now recognized that
reality and have introduced Community Groups to facilitate
specification development in areas of rapid innovation which are too
early for standardization.
And I agree Danny, that as a spec matures and stabilizes, it is quite
useful for the ecosystem to converge to a fixed point - a versioned
spec. And to be sure, there will continue to be innovation beyond
which would lead to the next version.
For convergence you typically need to specification to evolve as well.
Because with convergence you typically get better understanding of the
problem space and holes in the specification.
(And then parts of specifications become obsolete if we later decide
on an alternative way of doing something, and it would be beneficial
if specifications were updated directly to reflect that new direction
so people do not end up implementing the wrong thing. E.g. that can
happen now (and does happen) with HTML4, DOM1*, DOM2*, DOM3*, CSS1, etc.)
At TPAC this year, the agenda is being driven by participants who will
be proposing brainstorming sessions of various types. Let me know if
you want to lead a discussion on how we get better on "obsoleting"
existing specs (or parts thereof).
