> On Sep 18, Joseph Carter wrote: > > It's a problem if there's no transition to speak of. We apparently have > > decided not to make policy that makes a bunch of packages instantly non- > > compliant without a reasonable transition.
On Sat, Sep 18, 1999 at 11:24:13PM -0500, Chris Lawrence wrote: > I think we're starting to bastardize the concept of "policy > compliance" here. No. This is simple release management. You don't go from proposal to final release without some kind of intermediate period which involves testing. And, as Policy defines interfaces -- interfaces being one of the most sacred aspects of a computer system -- it's very reasonable to be thorough when changing it. You simply have to give policy a chance in the real world -- where people can implement it and see some of the unanticipated implications -- before it's at all reasonable to make that policy mandantory. Depreciating existing practice bastardizing the concept of policy compliance because it's not abrupt enough? What problems would *that* approach solve? [Aside from gratifying some wierd sort of concept of absolute authority.] -- Raul
