removing libdb 4.3
I completely agree that libdb4.3 is ancient grot, and it should be
removed from the archive.
I am distressed that the maintainers decided to wait until the freeze to
do that. This is entirely *backwards*. The time to decide, "hey, this
library should be removed" is at the *beginning* of the release cycle,
not at the end.
The result is destabilization; right when we should not be
destabilizing, packages are being asked to adapt to a new situation and
potentially new bugs will appear as a result.
Can I plead that, in the future, we decide that a *freeze* applies to
package removals just as it does to package changes--and that the same
rules apply, namely, a package is only removed during a freeze if this
is necessary to fix an RC bug?
If the decision were to upgrade libdb4.3 to libdb4.6 without a change in
package name--if this were just a new upstream version which claimed
compatibility--the rules of the freeze would say "no way, too late, we
don't do that, especially for libraries, especially for libraries used
by a large number of other packages."
But if it gets a new number, then hey, suddenly all bets are off. And
the effects are *exactly the same*: namely that the interaction with the
new library and existing software which adapts to it are untested.
I ask therefore that we create a new rule next time: maintainers which
wish to ditch old versions of libraries--and this is certainly an old
version that needs to be ditched, indeed, 4.4 too I would think--do so
at the BEGINNING of the release cycle, and that once the freeze has
happened, it's too late to start ditching things you have suddenly woken
We have a problem all the time with maintainers who want to do big
destabilizing things at the end of the release cycle, because they
suddenly see that the clock is ticking. The same strict attitude needs
to apply to ditching old libraries as it does to the *less*
destabilizing upgrade of a "leaf" application package.