Re: Please respect Freeze Policy
On 03/12/14 15:36, Sam Hartman wrote:
> You assert that even for leaf packages we should not make changes
> in unstable targeted beyond jessie. I can see an argument for that
> prior to the important bug deadline. However, past that point, I don't
> see value in avoiding changes to leaf packages in unstable, and I'd like
> to understand why you'd encourage folks to do so.
Avoiding non-testing-suitable changes (and particularly, new upstream
versions) in unstable gives you, as maintainer, more flexibility with
what you can get into testing in future, and gives potential NMUers the
same benefit.
testing-proposed-updates gets basically no testing, so the RT are
reluctant to use it for anything that is not an isolated and minimal RC
bug fix.
unstable has more chance of catching regressions before they hit
testing, so the RT are more willing to accept bugfixes whose risk/reward
ratio looks good enough if they can go via unstable. Note that the
deadline for pre-approved changes (which can potentially include
lower-severity bugs if the fix is sufficiently small and obviously
correct) is further away than the deadline for non-pre-approved
important bugfixes.
Uncoordinated new upstream versions are worse than uncoordinated Debian
patches or packaging changes, both because they're likely to be larger
and because if a patch or packaging change is a problem, it's typically
easy for the maintainer to revert it and make a new upload to unstable
whose diff relative to testing is smaller.
S
Reply to: