> >From BIND changelog:
> * Tue Jun 19 2007 Adam Tkac <atkac redhat com> 31:9.5.0a5-1
>
> It means BIND lives in rawhide/F8 about five months and I have
> reported only one more serious issue (#400461, that is why this thread
> was started). I believe that "a" word appended to version feels people
> with dread but I have to say not in this case. I'm ready to downgrade
> F8 BIND to latest stable but I can't see any argument why. One bug?
> Because it is alpha? These are not arguments whose could controvert my
> opinion that it was good decision put 9.5 to F8


Hello Adam,

I think an alpha release might be ok if we need to move forward
for certain features, but we should try hard to stay on a stable
release. So if that is possible, a released version makes much more
sense.

regards,

Florian La Roche

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wed, 2007-11-28 at 11:24 +0100, Adam Tkac wrote:
> On Tue, Nov 27, 2007 at 11:41:13PM +0100, Mark wrote:
> >
> > Hey,
> >
> > I understand why you choose to include it but i don't really like it.
> > If i run a server i want to have the latest _stable_ releases and
> > certainly no alpha (BIND) beta or even a build (NetworkManager)! now i
> > can imagine that desktop applications that aren't in a final release
> > can get pushed in Fedora because it simply has some nice additions
> > that you don't want your users to miss. But BIND is a vital part of a
> > server (DNS Server) so i think you shouldn't include beta's or even
> > alpha's of that in final releases of Fedora.
> >
> > And you say:
> > > Fedora has new features and new features mean bugs so you
> > > cannot expect such stability like RHEL.
> >
> > Oke i understand that. BUT the fact that you do push a vital server
> > component that is in alpha in Fedora does imply that you are testing
> > it on fedora for RHEL! (which in term keeps your other statement
> > standing).
> >
> > Fedora is used for servers (which you as a redhat employee probably
> > know) but in the mean time it's purpose is mainly a desktop OS. I
> > would say that:
> > - All server components like Sendmail, DNS, Apache, MySQL, PostgreSQL
> > etc.. should stay up to date with there latest _stable_ release (no
> > alpha's, beta's or rc's)
> > - All desktop related applications can probably be less tight.. I
> > don't see a problem there for alpha's, beta's or rc's.. as long as the
> > applications itself don't crash and just work.
> >
> > That's just my point of view as a fedora desktop and (previously) server user.
>
> >From BIND changelog:
> * Tue Jun 19 2007 Adam Tkac <atkac redhat com> 31:9.5.0a5-1
>
> It means BIND lives in rawhide/F8 about five months and I have
> reported only one more serious issue (#400461, that is why this thread
> was started). I believe that "a" word appended to version feels people
> with dread but I have to say not in this case. I'm ready to downgrade
> F8 BIND to latest stable but I can't see any argument why. One bug?
> Because it is alpha? These are not arguments whose could controvert my
> opinion that it was good decision put 9.5 to F8

I have to agree with you. The "alpha" name in version mostly doesn't
mean much. If you really tested it thoroughly I don't think you should
be blamed. Because otherwise nobody could allow for example Evolution
into the distro as it sometimes eats e-mails in my configuration (the
bug is reported upstream for a long time and no fix is ahead).

--
Tomas Mraz
No matter how far down the wrong road you've gone, turn back.
Turkish proverb

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

Adam Tkac wrote:

As I wrote above I disagree with you that put alpha (which will come
to beta really soon) to F8 is bad decision "because it is alpha". I
tested it and I didn't find any issues. That's why I decided put 9.5
to F8.

Adam


Is a former bind maintainer I have to support Adam here. In my opinion
it depends what package is taken and it's really a big difference
between a huge project (like gnome/gcc) and relative small one like bind.


Bind alpha/beta versions are periodically more stable than other
**stable** projects.


Anyway, do you know what rules have upstreams of all our packages? What
is a stable version in one can be an alpha for others. We can hardly
make strict general rules if reliable and balanced measure is missing.


Martin

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wed, 2007-11-28 at 07:33 +0100, Christopher Aillon wrote:
> On 11/28/2007 06:56 AM, Jason L Tibbitts III wrote:
> >>>>>> "TC" == Tom "spot" Callaway <Tom> writes:
> >
> > TC> Such as? Open to suggestions here.
> >
> > We had
> > http://fedoraproject.org/wiki/Extras/Schedule/MaintainerResponsibilityPolicy
> > which was never finished; the only thing in there is "Maintain
> > stability for users". I honestly don't see how you can be much more
> > specific without introducing needless bureaucracy. After all, the
> > alpha releases of some projects are more stable then the full releases
> > of others.
>
> This seems pretty much perfect, actually. What does it need in order to
> be "finished"?

Well, I'm not sure how it can be considered perfect when it does not
begin to address the "alpha/beta" issue that you think is resolvable
with packaging policy.

FWIW, I agree with Tibbs, since we have no way of determining how stable
package releases are without trusting the maintainer.

~spot

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wed, 28 Nov 2007 07:46:53 -0500
"Tom "spot" Callaway" <tcallawa@redhat.com> wrote:

> Well, I'm not sure how it can be considered perfect when it does not
> begin to address the "alpha/beta" issue that you think is resolvable
> with packaging policy.

"In general, it is preferred if maintainers avoid releasing "alpha" or
"beta" builds of packages into Fedora (both Rawhide and released
updates). However a package maintainer has the right to use their own
discretion regarding this issue and may provide whatever (s)he sees fit
for the user base. Things to consider include the amount of testing an
alpha or beta release has seen, the timeline to turn said alpha or beta
into a stable release, the feature sets provided, or the bugs fixed.
There may be other factors at play as well, which is why the person
best suited to make such a decision is the package maintainer in
question."

--
Jesse Keating
Fedora -- All my bits are free, are yours?
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wednesday 28 November 2007 07:55:34 Jesse Keating wrote:
> "In general, it is preferred if maintainers avoid releasing "alpha" or
> "beta" builds of packages into Fedora (both Rawhide and released
> updates).

How about unreleased cvs snapshots? kdepim in F7 failed to work after a recent
upgrade (and had been working fine) and my problem was only solved by
upgrading to F8 which had a more recent cvs snapshot that is still buggy but
usable. Rolling back did not work. The only support I got by the Fedora
maintainer was telling me to report it upstream. Upstream does not support it
because its a cvs snapshot and not a released version.

What do we do about that? This is the current status of kdepim in both F7 & 8.

-Steve

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wed, 2007-11-28 at 08:48 -0500, Steve Grubb wrote:
> On Wednesday 28 November 2007 07:55:34 Jesse Keating wrote:
> > "In general, it is preferred if maintainers avoid releasing "alpha" or
> > "beta" builds of packages into Fedora (both Rawhide and released
> > updates).
>
> How about unreleased cvs snapshots?

Not just that, we should pull out cdparanoia, as it has been in alpha
since 2001. :P

~spot

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wed, 28 Nov 2007 08:48:24 -0500
Steve Grubb <sgrubb@redhat.com> wrote:

> How about unreleased cvs snapshots? kdepim in F7 failed to work after
> a recent upgrade (and had been working fine) and my problem was only
> solved by upgrading to F8 which had a more recent cvs snapshot that
> is still buggy but usable. Rolling back did not work. The only
> support I got by the Fedora maintainer was telling me to report it
> upstream. Upstream does not support it because its a cvs snapshot and
> not a released version.
>
> What do we do about that? This is the current status of kdepim in
> both F7 & 8.

This again falls under the Maintainer's discretion. They can choose to
backport a specific fix as a patch, or pick up a CVS snapshot and run
with it, in order to fix the bug(s) in question.

--
Jesse Keating
Fedora -- All my bits are free, are yours?
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On Wed, 28 Nov 2007 09:22:16 -0500
"Tom "spot" Callaway" <tcallawa@redhat.com> wrote:

> Not just that, we should pull out cdparanoia, as it has been in alpha
> since 2001. :P

Again, it's a general statement not a blanket policy. The maintainer
has the right to contradict the general statement when it makes
reasonable sense to.

--
Jesse Keating
Fedora -- All my bits are free, are yours?
--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list

On 11/28/2007 01:46 PM, Tom "spot" Callaway wrote:

On Wed, 2007-11-28 at 07:33 +0100, Christopher Aillon wrote:

On 11/28/2007 06:56 AM, Jason L Tibbitts III wrote:

"TC" == Tom "spot" Callaway <Tom> writes:

TC> Such as? Open to suggestions here.

We had
http://fedoraproject.org/wiki/Extras/Schedule/MaintainerResponsibilityPolicy
which was never finished; the only thing in there is "Maintain
stability for users". I honestly don't see how you can be much more
specific without introducing needless bureaucracy. After all, the
alpha releases of some projects are more stable then the full releases
of others.
This seems pretty much perfect, actually. What does it need in order to
be "finished"?


Well, I'm not sure how it can be considered perfect when it does not
begin to address the "alpha/beta" issue that you think is resolvable
with packaging policy.

FWIW, I agree with Tibbs, since we have no way of determining how stable
package releases are without trusting the maintainer.


I never said that we should resolve the alpha/beta issue. I said we
should have some form of (loose) criteria for maintainers in release
branches. "It must maintain stability" is a good criterion item. Else,
maintainers could just go breaking stuff and say "well, I thought Fedora
was supposed to be bleeding edge - nobody told me I couldn't break stuff
in a release".


It also serves as a great fallback policy in the unlikely case we
(FESCo? RelEng?) ever find ourselves in the case where we need to decide
to (nudge the maintainer to) revert a change because it breaks too many
people.


--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list