On Fri, Mar 12, 2010 at 01:26:26AM +0100, Kevin Kofler wrote:
> Chris Adams wrote:
> > Developers can't always get what they want. Just because Fedora updates
> > to upstream's release-of-the-day doesn't mean Ubuntu, SuSE, etc. have
> > updated (so hopefully upstream is still paying attention to older
> > releases).
>
> It is (especially for leaf packages) much more likely they're going to say
> one or more of:
> * "screw you, use a sane distro" (and we lose the user),
> * "just build our tarball from source" (with the result that the user ends
> up with an unpackaged mess in /usr/local which grows like mold),
> * "screw the distro packages, use ours" (and this leads us to the "third-
> party package chaos" problem, which is characterized by low-quality
> packages, dependency hell, inter-repository compatibility issues etc.).
>
> > Most reasonable upstreams I have worked with fully understand that not
> > everybody is running yesterday's release and will work with you if you
> > find a problem with an older release. If an upstream can't handle that,
> > I would say that is the upstream's problem, not Fedora's.
>
> Upstream cannot go back in time and magically fix a bug in an old release.
> The bug is often already fixed in the current release, so the solution is
> for us to package the current release. And no, backporting fixes is often
> not practical.
I know that backporting fixes is often not practical, especially if there is
a lot of code churn upstream. How often this is the case depends on the
project of course. Do you have any cost estimates on that though? (honest
question, I'm not involved with KDE so I can't even guess)
As in, on average what are the costs of leaving a bug in vs. the cost of
updating to a new release. I noticed that there's a number of bugs that only
affect a subset of users that (often) can work around the issue. So the cost
- while high to a particular user - may be quite limited.
Updating to a new release will affect _every_ user and given that software
is imperfect it's likely that existing bugs will just be replaced by new,
different bugs. So on the one side you have the cost of having known bugs,
on the other side you have the benefit of fixing (some) bugs but the cost of
new bugs and a potential change in the user experience. Which one is higher
than the other one?
Cheers,
Peter
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
Mathieu Bridon wrote:
> « Alright, today I'll be implementing feature XYZ in my Foobar program. »
>
> [... a short hack later, testing the change...]
>
> « Why doesn't it work? It was working fine last time I tried, what's
> happening »
>
> [... an hour of debugging later...]
>
> « $#@!%µ the Libbar librazy was updated and isn't compatible anymore.
> Great, I just lost an hour trying to solve a problem in my program
> when it was coming from an soname bump! Thank you (not) $distro
> maintainer!!! »
It takes you an hour to figure that out? The error message you get when
you're trying to use a library with the wrong soname is quite clear!
And the next time it happens, you know that you should just make clean; make
before checking anything else if something like this happens.
Kevin Kofler
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
On Thu, 2010-03-11 at 23:41 +0000, Matthew Garrett wrote:
> You don't see a problem with breaking someone's application just because
> they've installed an update to a stable release of Fedora? It's
> obviously fine to do so when upgrading between releases, but within a
> release it's just gratuitously irritating.
On F13, upgrade gnome-panel to version in updates-testing and you'll get
panel crashes almost immediately when tryign to start an application or
just logout of gnome. Soo, if we were using rawhide, almost every gnome
user would be having crashes soon as they upgrade (minus the ones that
watch for breakage first before they upgrade). Anyway, perfect example
to test first (cuz maintainer might have accidentily missed something),
and not just update straight to the branch.
--
Mike Chambers
Madisonville, KY
"Best lil town on Earth!"
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
On Thu, Mar 11, 2010 at 03:59:46PM -0500, Simo Sorce wrote:
> On Thu, 11 Mar 2010 14:56:05 -0500
> Konstantin Ryabitsev <icon@fedoraproject.org> wrote:
>
> > (And if the answer is "backport the security fixes to 1.8.1" then I'm
> > afraid I don't really have the skills nor have the time to spend on
> > such massive effort).
>
> You can always find a co-maintainer skilled enough to help you in such
> rare cases... just saying.
>
The Fedora Objectives page[1] says: "In general, we prefer to move to a
newer version for updates rather than backport fixes". I know this is
what we're talking about, but it's not like we lack existing policy on
this - the idea that there's no direction to help maintainers do the
same things is not correct.
Ewan
[1] https://fedoraproject.org/wiki/Objectives
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
>> Is ARPA expecting everyone to upgrade to a sha256 supporting bind
>> immediately? There's no migration window?
>
> If someone has dnssec enabled in bind including DLV, then the key will be
> found and its use will be attempted. I am not sure what happens on an older
> bind 9.6.1 when that happens. One will hope it will just continue to be
> treated as "insecure" and not as "bogus" (aka servfail). I have not tested
> this.
Just for the record, 9.6.1 was patched so unknown algs go "insecure", so this
is not an issue. Sorry to distract from the main focus of this discussion with
a bad example.
Paul
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
On Fri, Mar 12, 2010 at 01:15:56AM +0100, Kevin Kofler wrote:
> Matthew Garrett wrote:
> > You don't see a problem with breaking someone's application just because
> > they've installed an update to a stable release of Fedora? It's
> > obviously fine to do so when upgrading between releases, but within a
> > release it's just gratuitously irritating.
>
> The application is not broken if whoever built it does his/her job. It's not
> like there's no notification about soname bumps (when done right; I know
> some people do not follow procedures, but then that's the problem, not the
> soname bump).
If the software is not maintained within Fedora, there's no notification
of soname bumps.
--
Matthew Garrett | mjg59@srcf.ucam.org
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
Dne 12.3.2010 02:24, Rahul Sundaram napsal(a):
> I disagree. Imagining that we are living in a island where no software
> exists outside the repository is just delusional and the assumption that
> everyone has the bandwidth to deal with all that churn is wrong as
> well. I should make people sit in a dial-up connection and have them
> update software now and then to bring them back to the ground.
Rahul, I used Debian, I know Debian, Debian was a friend of mine. I
don't want Fedora to be just yet another copycat of Debian. Please keep
it Fedora instead.
He uses statistics as a drunken man uses lamp-posts... for
support, rather than illumination.
-- Andrew Lang
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
Dne 12.3.2010 02:24, Rahul Sundaram napsal(a):
> I disagree. Imagining that we are living in a island where no software
> exists outside the repository is just delusional and the assumption that
> everyone has the bandwidth to deal with all that churn is wrong as
> well. I should make people sit in a dial-up connection and have them
> update software now and then to bring them back to the ground.
Rahul, I used Debian, I know Debian, Debian was a friend of mine. I
don't want Fedora to be just yet another copycat of Debian. Please keep
it Fedora instead.
Dne 12.3.2010 02:26, Mike Chambers napsal(a):
> On F13, upgrade gnome-panel to version in updates-testing and you'll get
When was F13 released? Oh, it wasn't, so it is just glorified Rawhide
still? And you complain about it being broken (especially in its
updates-*testing*)?
BTW, this is the reason why I disagree with the way how we did unfrozen
Rawhide. Instead of getting even more free version of Rawhide, we've got
one more make update and similar non-sense.
He uses statistics as a drunken man uses lamp-posts... for
support, rather than illumination.
-- Andrew Lang
--
devel mailing list
devel@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel
Stable Release Updates types proposal (was Fedora Board Meeting Recap 2010-03-11)
