From: Andrew Gierth (andrew@erlenstar.demon.co.uk)
Date: Thu May 09 2002 - 00:30:13 CDT
>>>>> "Henry" == Henry Spencer <henry@spsystems.net> writes:
Henry> Note that the entry point of the standards track, Proposed
Henry> Standard, explicitly does not require implementation
Henry> experience (although it is considered desirable...
That doesn't mean that a proposed standard has carte-blanche to
recommend or require new features which are known to cause problems
with existing deployed systems.
It's specifically a Bad Idea to publish a standard (and the fact that
it is only 'proposed' makes no difference in this case) which is such
that an implementor, following the standard carefully, will produce an
implementation which would then require substantial changes to make it
usable with existing software.
Henry> but note that *most* of our draft does have such experience
Henry> behind it).
_Most_ of our draft, considering only the technical issues rather than
the policy ones, is just existing implementation and therefore not
controversial.
Not enough attention is being paid to the lessons learned from
son-of-1036, where virtually every "new" feature had to be removed
either because it was never significantly implemented (like
Article-Names) or because it was actively dangerous (like
Also-Control).
-- Andrew.