Proposed udpates policy change

mike cloaked mike.cloaked at gmail.com
Wed Mar 10 22:14:44 UTC 2010


On Wed, Mar 10, 2010 at 9:11 PM, Gilboa Davara <gilboad at gmail.com> wrote:

>> If Fesco is aiming at getting rid of all the pesky packagers maintaining low
>> profile packages: You're well on your way.
>
> I usually stay away from mega-threads, but well put!
>
> I doubt that even major bug fixes in any of my (small) packages, ever
> got more than 1-2 karma votes. Many got zero - not even a vote by the
> original bug report owner!
>
> Why am I getting punished because some package didn't get enough testing
> (due to the low visibility of update-testing?) before it was pushed into
> -updates and caused breakage?
>
> Either we (package maintainers) are qualified to make sane decisions
> about our package or we are not. I don't really see a middle ground
> here.

I thought I would add a few thoughts to this now long running thread.

Firstly I have been a long standing - since Fedora Core 1 - user of
Fedora, and in general Fedora Linux has served me well through many
generations of new installs, across not only my own machines but also
those of relatives/friends whom I have been trusted to convert to sole
Linux use, up till now very successfully.

There have been large changes in recent times (KDE 3.5->4, major
changes to graphics drivers including open source Radeon/nouveau,
major boot process updates, KMS, pulseaudio etc etc). We have survived
all of those largely unscathed.  I would hope that the machines that I
run on behalf of other users will continue to serve them well through
yum updates whilst in normal production service.  I am lucky that I
can run my main machines on a released and current version of Fedora
that I expect will not fail in a catastrophic way after normal updates
- but I do have available other non-critical machines on which I can
run alpha or beta ( or even rawhide occasionally) versions, or run
current releases but be prepared to take the risk of running
unreleased packages from koji before they even hit bodhi, and before
they reach testing repos.  If these machines suffer in a major way it
is not a disaster and I can re-install at worst - but the main
production machines remain up and running (until recently that is!)  I
am lucky since I can usually find what fix is needed and sort it out.
However Aunt Bessie can't and relies on people like me to fix their
machines when their email stops working and a message appears on their
screen saying something that is incomprehensible to her!

Fine if it is only Aunt Bessie that I have to fix - but if Uncle Bob,
Grandma Celine, Grandad David, and 25 other assorted friends, cousins
and relatives all find their email stops one morning then I am going
to be unable to do my dayjob if I spend all my time getting their
machines all fixed because an update broke their production systems.
At that level I would say that the update that caused that level of
failure is no longer acceptable as a released update.  I know that we
are all human, and that occasionally we will all make a mistake (I do
too!) but there is a threshold beyond which a failure is really not
acceptable and once crossed there is a chance that users and testers
will be alienated and move elsewhere. In that event I think the
responsible person(s) should gracefully accept that a threshold has
been crossed and learn from flack that ensues, even that has arisen
from an upstream change that they were largely unaware would have
serious consequences.

Remember also that there are users who only have a single machine and
if that breaks then it is much harder for him/her to sort out the
problem if there is a loss of email functionality and/or loss of dns
(remember the dnssec issue) leading to loss of network connectivity
since getting the information required to fix it will need access to
the net - so critical packages do need to be identified and tested at
a more intense level than less critical packages. The kernel is also
clearly critical and when dependencies on X and graphics drivers could
break machines then that needs special consideration also.

In general packages and their maintainers do a good job and we get
regular excellent updated packages almost daily - what a service! -
users of other alternatives to linux certainly don't get that level of
provision or anything like it!

I know that there has been a lot of soul searching and a genuine
attempt to move forward - let's all keep level headed and try to be
constructive rather than destructive in trying to make for a better
Fedora. We have support that is truly up to date - let's keep it that
way, but also avoid really serious breakage on production releases.

-- 
mike c


More information about the devel mailing list