FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor


 
 
LinkBack Thread Tools
 
Old 10-12-2008, 09:33 PM
"D. Hugh Redelmeier"
 
Default

| From: Les Mikesell <lesmikesell@gmail.com>

| But the first question should be why a separate community is necessary. Why
| is it not possible for one of fedora's goals to be to provide a clean
| transition to RHEL or Centos at the end of certain development cycles

I think that this idea has some merit. I'm taking this out of context
to use as a jump-off point.

[I'm a long-term Red Hat and Fedora user. I also use Ubuntu
sometimes.]

Ubuntu 8.04 LTS has seduced me with its simultaneous promises of being
maintained and having a current code base.

RHEL/CentOS feels too old. Concrete issues:

- support for video cards

- PostgreSQL 8.3 vs 8.1

Ubuntu 8.04 will be quite stale before the next LTS comes out.
Probably more stale that RHEL/CentOS. It is all a matter of where
each stream is in the cycle. Right now, Ubuntu LTS is ahead of
RHEL/CentOS.

Ubuntu LTS is in the same series as regular Ubuntu. RHEL/CentOS are
not the same as Fedora. One could predict that the transition costs
between Ubuntus are lower that the transition costs from RHEL/CentOS
to or from Fedora. This is a disadvantage.

I find the various Fedora-targeted 3rd party repositories confusing
and even conflicting. This isn't healthy. I've not had this problem
with Ubuntu but I'm not sure why.


I like that fedora is willing to take radical steps. This is the only
way to do some important experiments.


Some changes make me (a conservative at heart) uncomfortable:

- the idea that network connectivity is part of a session just seems
bizarre to me. Network Manager may be useful but I think that this
aspect does not fit in with my UNIX world-view.

- the idea that non-X consoles will go bothers me. Just a few minutes
ago, I used a non-X console to whack on an X problem. I hit a lot
of X problems due to the way hardware gets introduced more quickly
than X versions are released (and debugged). I may stop using
Fedora when this change comes to pass.

- Documentation that I expect is not provided.

- the anatomy of the system changes more quickly than my
understanding. HAL/d-bus/etc. all seem like magic to me.



Why should a (source) package release be tied to a distro release?
For a lot of packages, the actual minimum requirements on the
environment are satisfied by earlier distro releases.

The strongest argument against is to do with testing: real
dependencies might diverge from declared dependencies and only testing
can show this. This adds a lot of testing burden to the package
maintainer.

Some packages really are not independent modules. Updating such a
package may require updating a lot of others. At least sometimes
these binary packages are all created from a single source package.


I wonder if smooth evolution (frequent uncoordinated package updates),
as opposed to punctuated equilibrium (updates tied to distro releases)
would work.

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 09:38 PM
Kevin Kofler
 
Default

Les Mikesell <lesmikesell <at> gmail.com> writes:
> Again, that seems like bad planning. Wasn't that predictable?

It was predictable, but not avoidable.

> But, there are people who claim to have don't successful upgrades from FC6 to
> RHEL/Centos.
(I assume you mean "have done", not "have don't" which makes no sense.)

There are people who *claim* many things, I doubt their claims make sense,
because really the first (and likely the last) RHEL which FC6 will have even
the slightest chance of safely upgrading to will be RHEL 6. And that will be an
upgrade skipping many Fedora releases and thus really bumpy. And at that point
you'll have had a support hole of somewhere around 2 years!

Going from FC6 to RHEL 5 is essentially a downgrade, especially so if you got
the numerous upgrades which went out after the FC6 release (if you never
updated anything, not even security stuff, then you might have more success,
but Fedora security upgrades are often new versions, RHEL security upgrades are
usually backports) and the infrastructure doesn't support downgrades at all,
neither Anaconda nor YUM support them.

Kevin Kofler

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 09:55 PM
Kevin Kofler
 
Default

Les Mikesell <lesmikesell <at> gmail.com> writes:
> user's perspective the approach was always wrong anyway. It would be
> good for both RHEL and fedora if they didn't really branch within a
> version - that is, at some point the RH betas take over the
> corresponding fedora release and fedora takes its wild and crazy changes
> to a new release.

Look, what you're asking for is really impossible:

1. There is no way RHEL will branch from a near-EOL or EOL Fedora release just
so you can safely upgrade from that version to RHEL. It would mean throwing out
at least a year of development and going back to old stuff, and RHEL would ship
with very outdated software (there are *already* complaints about its software
being outdated, but that's unavoidable due to stabilization/QA and
certification, but users would not tolerate even older software, they'd use a
competitor with newer software). It would also not help those who already
upgraded to a newer Fedora release.

2. There is also no way Fedora will stop pushing version upgrades to the
releases RHEL gets branched from, that's not how Fedora works. Fedora is about
bringing current software to our users and version upgrades are part of that.
There are also several cases where backporting security fixes is much harder
than just upgrading to a new version. It would also affect not just the version
RHEL gets branched from, but the previous version (or even 2 versions below if
it's the first month after a release) too, for obvious upgrade path reasons.
And many Fedora users would see a move like that as RH perverting Fedora for
their own purposes.

So Fedora will always have at least some packages which are newer than the RHEL
versions at any point in time, and thus it is impossible to guarantee an
upgrade path in that direction.

> Even so, some people claim to have done successful upgrades from fedora
> to the corresponding EL/Centos with just a few quirks that probably
> could have been avoided with a little central planning.

No amount of planning is going to avoid the fundamental impossibility outlined
above. The upgrades you mention were either from long-EOL Fedora releases (that
can work out, but it means months or even years without security updates!) or
involved downgrading several packages (e.g. the kernel).

> That is, even if it is impossible to coordinate the changes that lead to
> RHEL/Centos, there should be a way to have the final fedora update do a clean
> conversion - or at least better than most users could do on their own.

Even if you think there should be a way, the fact is that there is none.

Kevin Kofler

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 10:00 PM
Kevin Kofler
 
Default

Les Mikesell <lesmikesell <at> gmail.com> writes:
> With a planned progression to an enterprise version, that would not
> really be a migration away from fedora but the expected end point where
> you are permitted to continue using anything you've contributed or
> developed for your own use, staying in the same community instead of
> having all previous work dumped out the window at the end of a cycle.

You don't have to throw away the development you've done, you just have to keep
it updated with progress in technology. Or find somebody else who will, e.g. by
getting your work integrated into some upstream project or by starting a Free
Software project on your own. Fedora is not and will never be about "fire and
forget" type of development.

Kevin Kofler

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 10:00 PM
Kevin Kofler
 
Default

Chuck Anderson <cra <at> WPI.EDU> writes:
> So we should make sure Fedora can upgrade old RHEL installations?

That actually mostly works. "upgradeany" is your friend.

Kevin Kofler

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 10:37 PM
Emmanuel Seyman
 
Default

* Les Mikesell [12/10/2008 22:14] :
>
> Yes but what is the point of developing for it?

There's a notion called freedom that you may have heard of.

> With a planned progression to an enterprise version, that would not
> really be a migration away from fedora but the expected end point where

Again, this supposes that one of Fedora's end goals is to easily
permit its users to migrate to other distributions. This isn't the
case.

> you are permitted to continue using anything you've contributed or
> developed for your own use, staying in the same community instead of
> having all previous work dumped out the window at the end of a cycle.

When a Fedora version release reaches EOL, users have the possibility
of upgrading to the next release. These days, they even have the option
of upgrading to the release following that one if they wait long enough.
I have no idea where you get the notion that we're guiding users to a
"planned dead-end" or that, once a Fedora release is EOL-ed, they have
to dump out their work out the window.

If your goal is to use an distribution that promises ABI/API
compatibility, long term support and other "enterprise" features, there
are a whole host of distributions for which these are goals.
Why not use them ?

Emmanuel

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 10:37 PM
"Horst H. von Brand"
 
Default

Les Mikesell <lesmikesell@gmail.com> wrote:
> Patrice Dumas wrote:

[...]

> > It is a bit more complicated. A distro may begin its life bleeding
> > edge and become stable as time goes by, if it is still maintained. And
> > a stable distro may have parts that are bleeding-edge. This is not
> > necessarily easy to implement, but these scenarios certainly have
> > merits.

I don't see any... Sure, it might be nice for the (tiny minority) who wants
e.g. bleeding edge compiler (because they are hacking on it?) with a very
stable everything else.

> That was the way things worked when redhat developed its popularity.


No...

> An X.0 release was approximately as unstable as a fedora,

Not really.

> but as it
> updated to X.2 or X.3 it would have become very stable and people who
> started a development cycle with the early versions could keep the
> same OS in production as it matured.

But there was no guarantee whatsoever of a .1, a .2 or even a .3 would ever
materialize.

> What we need

Who is the "we" here?

> for the same effect
> now is for the versions of fedora that provide the initial RHEL cuts
> to offer a seamless update to the subsequent matching CentOS,
> repointing to its update repositories for continued support.

If a majority of Fedora folks wanted that, Fedora would move like
that. Fedora Legacy wouldn't have gone belly up for lack of hands. And Red
Hat would have gone on doing .0, .1 and so on forever.
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 2654431
Universidad Tecnica Federico Santa Maria +56 32 2654239
Casilla 110-V, Valparaiso, Chile 2340000 Fax: +56 32 2797513

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 10:40 PM
"Horst H. von Brand"
 
Default

Les Mikesell <lesmikesell@gmail.com> wrote:

[...]

> Fedora could make it's next release somewhere around the point where
> the paths start to diverge so people who wanted the fast-track
> unstable flavor could re-install as they apparently love to do, and
> the rest of us could just drift into stability.

Why would any Fedoran want to "drift into stability"? That is a
contradiction in terms... If it was /so/ badly wanted as you claim, Fedora
Legacy would be alive and well, don't you think?
--
Dr. Horst H. von Brand User #22616 counter.li.org
Departamento de Informatica Fono: +56 32 2654431
Universidad Tecnica Federico Santa Maria +56 32 2654239
Casilla 110-V, Valparaiso, Chile 2340000 Fax: +56 32 2797513

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 10:44 PM
Patrice Dumas
 
Default

On Sun, Oct 12, 2008 at 07:40:59PM -0300, Horst H. von Brand wrote:
> Les Mikesell <lesmikesell@gmail.com> wrote:
>
> [...]
>
> > Fedora could make it's next release somewhere around the point where
> > the paths start to diverge so people who wanted the fast-track
> > unstable flavor could re-install as they apparently love to do, and
> > the rest of us could just drift into stability.
>
> Why would any Fedoran want to "drift into stability"? That is a
> contradiction in terms... If it was /so/ badly wanted as you claim, Fedora
> Legacy would be alive and well, don't you think?

Things have changed since Fedora Legacy. Core has opened and
infrastructure has seen many improvements. What was not practical in the
old days is, in my opinion, possible now.

--
Pat

--
fedora-devel-list mailing list
fedora-devel-list@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list
 
Old 10-12-2008, 11:56 PM
Les Mikesell
 
Default

Horst H. von Brand wrote:

Les Mikesell <lesmikesell@gmail.com> wrote:

[...]


Fedora could make it's next release somewhere around the point where
the paths start to diverge so people who wanted the fast-track
unstable flavor could re-install as they apparently love to do, and
the rest of us could just drift into stability.


Why would any Fedoran want to "drift into stability"? That is a
contradiction in terms... If it was /so/ badly wanted as you claim, Fedora
Legacy would be alive and well, don't you think?


No, Fedora, legacy or not, is not good at maintaining stability. I'm not
surprised it didn't work and wouldn't expect it to work if revived.
What people actually do is run RHEL or Centos for their actual work.
Which leaves the question of how to get from one to the other as you
develop something new, then want to run it.


--
Les Mikesell
lesmikesell@gmail.com

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

Thread Tools




All times are GMT. The time now is 07:03 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org