9:40 p.m.
Hi,
I'm seeing 2.6.29 packages for f10 showing up on koji for a while now but
they never end up as a real update. Is there any particular reason for that?
Regards,
Dennis
5:15 a.m.
On Mon, 27 Apr 2009, Dennis J. wrote:
I'm seeing 2.6.29 packages for f10 showing up on koji for a while
now but
they never end up as a real update. Is there any particular reason for that?
I imagine that the intention is to go to 2.6.29 on f10 at some point, but
there are still issues to fix before it is ready for general release. For
example one such issue that is now solved is that 2.6.29 uses a different
version of squashfs to the one in f10, which meant that things such as
liveCDs would cease to work (this was solved by adding the code for the
earlier version of squashfs to 2.6.29), but there may be other things that
still need to be resolved.
Michael Young
5:30 a.m.
Dennis J. wrote:
Hi,
I'm seeing 2.6.29 packages for f10 showing up on koji for a while now
but they never end up as a real update. Is there any particular reason
for that?
Regards,
Dennis
It is already present in updates-testing (2.6.29.1-30), and 2.6.29-1.42
is heading to go there soon too:
https://admin.fedoraproject.org/updates/kernel-2.6.29.1-42.fc10
Regards,
Milos
9:46 a.m.
On 04/27/2009 12:15 PM, M A Young wrote:
On Mon, 27 Apr 2009, Dennis J. wrote:
> I'm seeing 2.6.29 packages for f10 showing up on koji for a while now
> but they never end up as a real update. Is there any particular reason
> for that?
I imagine that the intention is to go to 2.6.29 on f10 at some point,
but there are still issues to fix before it is ready for general
release. For example one such issue that is now solved is that 2.6.29
uses a different version of squashfs to the one in f10, which meant that
things such as liveCDs would cease to work (this was solved by adding
the code for the earlier version of squashfs to 2.6.29), but there may
be other things that still need to be resolved.
I guess I'm wondering why so much effort is put into this when by the time
the kernel gets released for f10 it is already outdated and we are moving
towards a f12 release which will end the support for f10 anyway.
If f10 and/or kernel 2.6.29 is going to be the baseline for RHEL 6 then I
could see why such an effort is made but other than that, why bother?
Regards,
Dennis
10:15 a.m.
On Monday 27 April 2009 16:46:16 Dennis J. wrote:
On 04/27/2009 12:15 PM, M A Young wrote:
> On Mon, 27 Apr 2009, Dennis J. wrote:
>> I'm seeing 2.6.29 packages for f10 showing up on koji for a while now
>> but they never end up as a real update. Is there any particular reason
>> for that?
>
> I imagine that the intention is to go to 2.6.29 on f10 at some point,
> but there are still issues to fix before it is ready for general
> release. For example one such issue that is now solved is that 2.6.29
> uses a different version of squashfs to the one in f10, which meant that
> things such as liveCDs would cease to work (this was solved by adding
> the code for the earlier version of squashfs to 2.6.29), but there may
> be other things that still need to be resolved.
I guess I'm wondering why so much effort is put into this when by the time
the kernel gets released for f10 it is already outdated and we are moving
towards a f12 release which will end the support for f10 anyway.
correction:
after F11 release and about two months, F9 won't be supported (F10 will)
If f10 and/or kernel 2.6.29 is going to be the baseline for RHEL 6
then I
could see why such an effort is made but other than that, why bother?
Regards,
Dennis
11:13 a.m.
On Mon, 27 Apr 2009, Dennis J. wrote:
I guess I'm wondering why so much effort is put into this when by
the time
the kernel gets released for f10 it is already outdated and we are moving
towards a f12 release which will end the support for f10 anyway.
If f10 and/or kernel 2.6.29 is going to be the baseline for RHEL 6 then I
could see why such an effort is made but other than that, why bother?
f10 has still 7 or 8 months to run. f9 is the version that will be
obsoleted soon, and that isn't being updated to 2.6.29. There is a good
chance that 2.6.30 will be out before f12 is released.
Michael Young
2:48 p.m.
On 04/27/2009 06:13 PM, M A Young wrote:
On Mon, 27 Apr 2009, Dennis J. wrote:
> I guess I'm wondering why so much effort is put into this when by the
> time the kernel gets released for f10 it is already outdated and we
> are moving towards a f12 release which will end the support for f10
> anyway.
>
> If f10 and/or kernel 2.6.29 is going to be the baseline for RHEL 6
> then I could see why such an effort is made but other than that, why
> bother?
f10 has still 7 or 8 months to run. f9 is the version that will be
obsoleted soon, and that isn't being updated to 2.6.29. There is a good
chance that 2.6.30 will be out before f12 is released.
That's really my point. 2.6.30 will be out soon so 2.6.29 will be outdated
again. What is the motivation of putting this much work into a kernel that
will be outdated when released and only have a shelf-life of 7 or 8 months.
Is there some critical feature in there that makes 2.6.29 so much better
than 2.6.27 to warrant this much effort for so little gain?
Regards,
Dennis
3:42 p.m.
On Mon, 27 Apr 2009, Dennis J. wrote:
That's really my point. 2.6.30 will be out soon so 2.6.29 will be
outdated
again. What is the motivation of putting this much work into a kernel that
will be outdated when released and only have a shelf-life of 7 or 8 months.
Is there some critical feature in there that makes 2.6.29 so much better than
2.6.27 to warrant this much effort for so little gain?
By that argument you would never update at all, because there will always
be a new kernel coming along in a few months. There is also a cost to
standing still, because things go out of date, and they would have to work
to back-port patches for security fixes etc.
Michael Young
3:46 p.m.
-------- Original Message --------
Subject: Re: 2.6.29 kernel for f10
From: Dennis J. <dennisml(a)conversis.de>
To: Development discussions related to Fedora <fedora-devel-list(a)redhat.com>
Date: 04/27/2009 02:48 PM
That's really my point. 2.6.30 will be out soon so 2.6.29 will be
outdated again. What is the motivation of putting this much work into a
kernel that will be outdated when released and only have a shelf-life of
7 or 8 months. Is there some critical feature in there that makes 2.6.29
so much better than 2.6.27 to warrant this much effort for so little gain?
2.6.29 works great on F10 for me and many others. Waiting for 2.6.30 is
not required and will waste *more* time.
I need 2.6.29 on my laptop so that suspend works. That's kind of important.
4:53 p.m.
Dennis J. wrote:
That's really my point. 2.6.30 will be out soon so 2.6.29 will be
outdated again. What is the motivation of putting this much work into a
kernel that will be outdated when released and only have a shelf-life of
7 or 8 months. Is there some critical feature in there that makes 2.6.29
so much better than 2.6.27 to warrant this much effort for so little gain?
Not being involved in the work involved for 2.6.29, but totally enjoying
it. It fixes a number of issues I've had on my system with various
hardware (USB, graphics...)
--
Nathanael d. Noblet
T: 403.875.4613
7:12 p.m.
On 04/27/2009 10:42 PM, M A Young wrote:
On Mon, 27 Apr 2009, Dennis J. wrote:
> That's really my point. 2.6.30 will be out soon so 2.6.29 will be
> outdated again. What is the motivation of putting this much work into
> a kernel that will be outdated when released and only have a
> shelf-life of 7 or 8 months. Is there some critical feature in there
> that makes 2.6.29 so much better than 2.6.27 to warrant this much
> effort for so little gain?
By that argument you would never update at all, because there will
always be a new kernel coming along in a few months. There is also a
cost to standing still, because things go out of date, and they would
have to work to back-port patches for security fixes etc.
That makes sense for a long lived distribution but not really for fedora.
If I really need to be running the latest versions then I'll update to the
next version of fedora when it comes out which is always a maximum of 6
months away. I'm certain there are some fixes in the new kernel that some
people will appreciate and there is certainly no harm in getting them but
actually spending all this time on apparently rather complex problems
between these two kernel versions seems strange if the result will be that
short-lived.
I'm mean 2.6.27 isn't the freshest kernel but it doesn't look that bad to
me that people couldn't wait for the release of f11 which is right around
the corner.
Regards,
Dennis
7:56 p.m.
Dennis J. wrote:
That makes sense for a long lived distribution but not really for
fedora. If I really need to be running the latest versions then I'll
update to the next version of fedora when it comes out which is always a
maximum of 6 months away. I'm certain there are some fixes in the new
kernel that some people will appreciate and there is certainly no harm
in getting them but actually spending all this time on apparently rather
complex problems between these two kernel versions seems strange if the
result will be that short-lived.
Just out of curiosity, since you aren't the one expending the effort,
why does it bother you that someone else is?
--
Nathanael d. Noblet
T: 403.875.4613
3:11 a.m.
On 04/28/2009 02:56 AM, Nathanael D. Noblet wrote:
Dennis J. wrote:
> That makes sense for a long lived distribution but not really for
> fedora. If I really need to be running the latest versions then I'll
> update to the next version of fedora when it comes out which is always
> a maximum of 6 months away. I'm certain there are some fixes in the
> new kernel that some people will appreciate and there is certainly no
> harm in getting them but actually spending all this time on apparently
> rather complex problems between these two kernel versions seems
> strange if the result will be that short-lived.
Just out of curiosity, since you aren't the one expending the effort,
why does it bother you that someone else is?
Because of the regressions and breakage that this new kernel will
invariably introduce. Personally, I also think this energy would be
better spent on F-11/rawhide develompent...
8:28 a.m.
On 04/28/2009 02:56 AM, Nathanael D. Noblet wrote:
Dennis J. wrote:
> That makes sense for a long lived distribution but not really for
> fedora. If I really need to be running the latest versions then I'll
> update to the next version of fedora when it comes out which is always
> a maximum of 6 months away. I'm certain there are some fixes in the
> new kernel that some people will appreciate and there is certainly no
> harm in getting them but actually spending all this time on apparently
> rather complex problems between these two kernel versions seems
> strange if the result will be that short-lived.
Just out of curiosity, since you aren't the one expending the effort,
why does it bother you that someone else is?
I couldn't help but wonder if there is something particular going on that
I've missed. If for example the reason for this effort is that this kernel
is going to be the baseline for RHEL 6 then that's interesting information
for me as I can then take a closer look at that version and keep that in
mind when planning my next deployments.
Also if the reason is that there are a considerable number of difficult
changes in the new kernel then I might to choose holding updates of other
packages back when the new kernel gets out so I can pinpoint if any
problems from that update are a result of the new kernel.
In general I find that if I come across developments that are in some form
out of the ordinary understanding *why* things happen that way often helps
me plan better for the future. Not that it really matter all that much in
this particular case. Like you I was just curious about what's going on.
Regards,
Dennis
9:42 a.m.
Once upon a time, Dennis J. <dennisml(a)conversis.de> said:
In general I find that if I come across developments that are in some
form
out of the ordinary understanding *why* things happen that way often helps
me plan better for the future. Not that it really matter all that much in
this particular case. Like you I was just curious about what's going on.
IIRC, the Fedora kernel team tries to keep all the current releases on a
similar kernel version, at least for a while. At this point, not much
energy will go into F9, since it will stop getting updates before long,
but F10 still has over half its life left, so updating it makes sense.
If they can keep F10 and F11 along similar lines, it cuts down their
work at tracking bugs and such (supporting one kernel version is enough
work; supporting two or three is a recipie for burnout).
The kernel of course has widespread impact, but as far as dependencies
go, it is relatively self-contained. New features may only be enabled
in newer Fedora to keep from impacting existing releases (i.e. some new
feature in 2.6.29 may only be enabled on F11 and not F10).
--
Chris Adams <cmadams(a)hiwaay.net>
Systems and Network Administrator - HiWAAY Internet Services
I don't speak for anybody but myself - that's enough trouble.
11:51 a.m.
Dennis J. wrote:
That makes sense for a long lived distribution but not really for
fedora.
If I really need to be running the latest versions then I'll update to the
next version of fedora when it comes out which is always a maximum of 6
months away. I'm certain there are some fixes in the new kernel that some
people will appreciate and there is certainly no harm in getting them but
actually spending all this time on apparently rather complex problems
between these two kernel versions seems strange if the result will be that
short-lived.
Fedora has always updated the kernel to upstream releases regularly. They
just decided to skip 2.6.28, which may be why you got out of the habit.
IMHO updating to new upstream releases is the right thing to do. New kernels
tend to fix lots of bugs. And FWIW, we're also handling KDE in a similar
way. IMHO more packages should get version upgrades. Sure, application
changes which remove features, library soname bumps which require
rebuilding half of the distro and stuff like that needs to go into the next
Fedora only, but new releases which don't introduce breakage, don't remove
features, fix many bugs and add new features definitely should get pushed.
In the case of the kernel, it's sometimes a bit of a tradeoff between a few
regressions which affect a few people and many bugfixes which affect many
more. The decision of pushing out a new kernel to stable is all about
finding the right point at which to push it, minimizing regressions while
at the same time not making people wait forever for their much needed
fixes.
Kevin Kofler
5478
days inactive
5479
days old
15 comments
9 participants
participants (9)
-
Chris Adams -
Denis Leroy -
Dennis Jacobfeuerborn -
Kevin Kofler -
M A Young -
Michael Cronenworth -
Michal Hlavinka -
Milos Jakubicek -
Nathanael D. Noblet