Jesus Climent <[EMAIL PROTECTED]> writes: > 2.20 became useless because spammers reacted to most of the rules developed, > since it did not have any learning engine. It needed new rules, new rules > needed new functions in the core code (not just rules updates) and new > functions needed new perl libraries with new versions. To use those libraries > you needed new core code, which basically meant having the whole 2.4x > backported.
This is an excellent argument for upgrading that part. It is no argument for including new command line features, hookins to other parts of the system, or other arbitrary features that might be added. So I have no objection to a policy which incorporates new virus definitions and the like, but not just arbitrary possibly destabilizing code. It might well be that it's a lot of work to provide the package rightly (where "rightly" means that it only upgrades what must be upgraded to keep it useful, but not other stuff). But that doesn't mean it's impossible, nor does it excuse doing it wrongly because it's too hard for the Debian package maintainer. We can get him help, or we can say that Debian doesn't have anyone who wants to provide a stable package. But "just provide an unstable package" is not a solution. We don't need the Debian brand to do that. Thomas
