testing versus unstable: tradeoffs?
People at my site wish to migrate away from Debian stable to
something more current. I understand that unstable is more bleeding
edge than testing, and that there is some kind of automatic process
whereby packages migrate from unstable to testing. (That automatic
process makes sure there are no (critical?) bugs filed against the
package, and I think that a package needs to ripen for some time in
unstable, before being allowed to migrate to testing.)
One can compare the distributions along several dimensions. Here are
some of them. I'd appreciate any comments you might have. Below, I'm
arguing in favor of unstable and against testing (as devil's advocate,
if you will). What do you think about those arguments? Prove me
* Number of bugs.
One might think that testing has fewer bugs than unstable. But the
truth is, since testing is an automatic process against filed bug
reports, bugs can always creep in there before they are discovered
and filed. And, what's more, problems will be fixed much quicker in
unstable than in testing, due to the ripening required in testing.
* Security fixes.
Security fixes come automatically with the new package installs in
unstable, whereas the security fixes department in the testing
distribution is not to be relied upon.