On Sun, Mar 07, 2004 at 01:55:24PM -0800, Thomas Bushnell, BSG wrote: > Anthony Towns <[EMAIL PROTECTED]> writes: > > There _is_ a change: one day we're distributing non-free, the next, > > we're not. That's the important change. It's not a change of policy, > > certainly, it's instead a claim that the *existing* policy does *not* > > need to be changed to meet the concern that Debian will always be > > distributing non-free software. > Except that there will probably always be non-free software, and even > if the amount on debian.org goes to zero, it might go positive the > next week.
That's quite possible. A plausible explanation for it might be an exponentially decreasing amount of useful non-free software, say N = 300 * 2^(-t/104) with an error of 50%, and t measured in weeks. Sure, once you get to week 752, you might actually drop to zero packages, then rise again a couple of weeks later because of the error factor. But with that trend, by the time you hit week 917, you're guaranteed to never have another non-free package again. I think it'd probably be reasonable to drop non-free at around week 650 when we're only going to be affecting a handful of packages, or possibly earlier, in the case, but the mere possibility of some fluctuation isn't a problem even if we decided to only remove non-free once we were confident there'd *never* be any useful non-free software needing packaging. > > > Really? What is it? What is the system for removing packages from > > > non-free? > > The maintainer says "this package is no longer needed" or "this packages > > has been relicensed under the GPL" or similar, and it gets removed. What > > did you think it was? > I believe this is an inadequate system. What do you think of a > compromise position which would allow a package in non-free only if > there is no free package filling the same niche? That's the system we've already got -- people don't like maintaining non-free software, so when there really is some free software that fills the same niche, it gets dropped by the maintainer. If you'd like to do QA work making sure that happens more promptly than it does atm, please do. Cheers, aj -- Anthony Towns <[EMAIL PROTECTED]> <http://azure.humbug.org.au/~aj/> I don't speak for anyone save myself. GPG signed mail preferred. Linux.conf.au 2004 -- Because we could. http://conf.linux.org.au/ -- Jan 12-17, 2004
signature.asc
Description: Digital signature