Important information regarding the merger of core and extras, and what this means to Legacy

[Jesse Keating]

See my blog regarding the future of Legacy as a project. Please remember
these are just proposals and not final solutions. A wiki page will follow
soon.

http://jkeating.livejournal.com/#entry_34659

[Eric Rostetter]

Post by Jesse Keating
See my blog regarding the future of Legacy as a project. Please remember
these are just proposals and not final solutions. A wiki page will follow
soon.

First I would like to say to those who say Fedora Legacy has failed, that
it _did_ work (i.e. didn't fail) for the most critical time period and the
most critical OS version (RHL 7-9, FC1). If it has failed, or is failing,
it must not be forgotten that before it failed it worked exceedingly well.

Second, I'm fairly comfortable with saying that if FC goes to a 13 month
support cycle, FL is basically not needed anymore. IMHO, people can upgrade
once a year when presented with a known/documented release cycle, and known
documented alternatives.

Now, I don't know that FC will change, or if FL is needed any more even
if FC doesn't change. But I do know that FL saved my life by being there
when I needed it, and while I don't really need it any more I'm forever
grateful to it for being there when I did need it.

[Jesse Keating]

Post by Eric Rostetter
First I would like to say to those who say Fedora Legacy has failed, that
it _did_ work (i.e. didn't fail) for the most critical time period and the
most critical OS version (RHL 7-9, FC1).  If it has failed, or is failing,
it must not be forgotten that before it failed it worked exceedingly well.

YES! I forgot to mention this in my blog, but when we were doing the RHL
updates, there was a LOT of interest and a lot of community participation,
and the project worked pretty darn well. I think it became evident by the
dramatic drop in participation that there just wasn't the interest we thought
there would be in long term Fedora releases. This is why I think its OK in
slightly extending the 'official' lifespan and bringing Legacy in to help
with that aspect.

Thank you for your feedback Eric, you've forever kept us honest! (:

[Matthew Miller]

Post by Eric Rostetter
First I would like to say to those who say Fedora Legacy has failed, that
it _did_ work (i.e. didn't fail) for the most critical time period and the
most critical OS version (RHL 7-9, FC1). If it has failed, or is failing,
it must not be forgotten that before it failed it worked exceedingly well.

Or at least moderately well. Let's not over-sell. :)

Post by Eric Rostetter
Second, I'm fairly comfortable with saying that if FC goes to a 13 month
support cycle, FL is basically not needed anymore. IMHO, people can upgrade
once a year when presented with a known/documented release cycle, and known
documented alternatives.

One month of annual overlap is still a bit short.

[Gene Heskett]

Post by Matthew Miller

Post by Eric Rostetter
First I would like to say to those who say Fedora Legacy has failed,
that it _did_ work (i.e. didn't fail) for the most critical time
period and the most critical OS version (RHL 7-9, FC1). If it has
failed, or is failing, it must not be forgotten that before it failed
it worked exceedingly well.

Or at least moderately well. Let's not over-sell. :)

Post by Eric Rostetter
Second, I'm fairly comfortable with saying that if FC goes to a 13
month support cycle, FL is basically not needed anymore. IMHO, people
can upgrade once a year when presented with a known/documented release
cycle, and known documented alternatives.

One month of annual overlap is still a bit short.

I can't help but agree that its too short. 3 or 6 would be much more
realistic from the users viewpoint, who has his setup all fine tuned and
doesn't want to go thru that on an annual basis. There are other things
to life you know.

[Jesse Keating]

Post by Gene Heskett
I can't help but agree that its too short. 3 or 6 would be much more
realistic from the users viewpoint, who has his setup all fine tuned and
doesn't want to go thru that on an annual basis.  There are other things
to life you know.

So why don't you use CentOS which as a annual or every other year release?

[Gene Heskett]

Post by Jesse Keating

Post by Gene Heskett
I can't help but agree that its too short. 3 or 6 would be much more
realistic from the users viewpoint, who has his setup all fine tuned
and doesn't want to go thru that on an annual basis.  There are other
things to life you know.

So why don't you use CentOS which as a annual or every other year release?

Theres several reasons, the old kernel version being one of them. Firewire
doesn't work that I know of, and I have a firewire movie camera.

[Jesse Keating]

Post by Gene Heskett
Theres several reasons, the old kernel version being one of them. Firewire
doesn't work that I know of, and I have a firewire movie camera.

And when CentOS5 comes out?

[Matthew Miller]

Post by Jesse Keating

Post by Gene Heskett
Theres several reasons, the old kernel version being one of them.

And when CentOS5 comes out?

Well, based on history, it'll be slightly behind-the-newest at release date
(RHEL stabilization + a month or so for CentOS) but generally current
enough, but then by this spring we'll see a batch of computers with hardware
that doesn't work.

[Jesse Keating]

Post by Matthew Miller
Well, based on history, it'll be slightly behind-the-newest at release date
(RHEL stabilization + a month or so for CentOS) but generally current
enough, but then by this spring we'll see a batch of computers with
hardware that doesn't work.

Isn't this where the quarterly updates with new hardware support come in?

[Matthew Miller]

Post by Jesse Keating

Post by Matthew Miller
Well, based on history, it'll be slightly behind-the-newest at release
date (RHEL stabilization + a month or so for CentOS) but generally
current enough, but then by this spring we'll see a batch of computers
with hardware that doesn't work.

Isn't this where the quarterly updates with new hardware support come in?

Is RHEL5 going to go wholesale to new kernel versions with the quarterly
updates, or is it actually going to backport all updated drivers to the
older release?

[Jesse Keating]

Post by Matthew Miller
Is RHEL5 going to go wholesale to new kernel versions with the quarterly
updates, or is it actually going to backport all updated drivers to the
older release?

From what I gather out in the community (not looking at any internal
documentation), that the new drivers would be backported.

[Matthew Miller]

Post by Jesse Keating

Post by Matthew Miller
Is RHEL5 going to go wholesale to new kernel versions with the quarterly
updates, or is it actually going to backport all updated drivers to the
older release?

From what I gather out in the community (not looking at any internal
documentation), that the new drivers would be backported.

We'll have to see how this works out.

[Eric Rostetter]

Post by Gene Heskett
Theres several reasons, the old kernel version being one of them. Firewire
doesn't work that I know of, and I have a firewire movie camera.

I missed what this is about, but if it is about the CentOS/RHEL kernel,
then simply install the unsupported kernel to get the firewire support.
Works for me at least (with firewire removable disk drives).

[Gene Heskett]

Post by Jesse Keating

Post by Gene Heskett
Theres several reasons, the old kernel version being one of them.
Firewire doesn't work that I know of, and I have a firewire movie
camera.

And when CentOS5 comes out?

I've no idea when, or if firewire is back among the living. It took till
the last new kernel for FC5 before it worked well enough to be usable.
Where does that place centos5 then?

[Jesse Keating]

I've no idea when, or if firewire is back among the living.  It took till
the last new kernel for FC5 before it worked well enough to be usable.  
Where does that place centos5 then?

RHEL5 kernel is largely based on the FC6 kernel.

[Gene Heskett]

Post by Jesse Keating

I've no idea when, or if firewire is back among the living.  It took
till the last new kernel for FC5 before it worked well enough to be
usable. Where does that place centos5 then?

RHEL5 kernel is largely based on the FC6 kernel.

Which isn't at all stable. Old kino, before a 1394 re-write was started
about 18 months ago now, was bulletproof and worked just fine on this
exact same hardware. It was very stable when controlling my camera. Now,
the two versions compatible with the later libraries etc are both so
unstable, doing segfault exits so quickly that the only chance I have of
using it is with the last FC5 kernel, running on my lappy. Either 8.0,
0.9.2 or 0.9.3 take a segfault exit, stage right, on about the second
mouse click, or even the first on this machine with the latest non-xen
kernel(s).

I don't think this is kino's fault, particularly since Dan is unable to
duplicate it on whatever system(s) he is using.

[Bill Perrotta]

That is fine where can i download an iso of centos? If it is similar enough to do all labs for rhel3 rhce that is my only concern.

Post by Gene Heskett
I can't help but agree that its too short. 3 or 6 would be much more
realistic from the users viewpoint, who has his setup all fine tuned and
doesn't want to go thru that on an annual basis. There are other things
to life you know.

So why don't you use CentOS which as a annual or every other year release?
--
Jesse Keating
Release Engineer: Fedora
--
fedora-legacy-list mailing list
fedora-legacy-***@redhat.com
https://www.redhat.com/mailman/listinfo/fedora-legacy-list


---------------------------------
Sponsored Link

Mortgage rates near 39yr lows. $510,000 Mortgage for $1,698/mo - Calculate new house payment

[Josep L. Guallar-Esteve]

Post by Bill Perrotta
That is fine where can i download an iso of centos? If it is similar enough
to do all labs for rhel3 rhce that is my only concern.

CentOS-3 is a rebuild of the freely available of source code of RHEL-3
CentOS-4 is a rebuild of the freely available of source code of RHEL-4

As a result of being a rebuild, Centos-X is binary-compatible with RHEL-X

You can download CentOS after a very easy search at the ultra-secret-website:
http://www.google.com/search?q=linux+centos


Regards,
Josep

[Nils Breunese (Lemonbit)]

Post by Bill Perrotta
That is fine where can i download an iso of centos? If it is
similar enough to do all labs for rhel3 rhce that is my only concern.

Come on, Google is your friend. Go to http://www.centos.org/ and take
it from there. CentOS 3 is exactly the same as RHEL 3, except for the
branding.

Nils.

[Jeff Sheltren]

Post by Bill Perrotta
That is fine where can i download an iso of centos? If it is
similar enough to do all labs for rhel3 rhce that is my only concern.

www.centos.org

CentOS is a rebuild of RHEL, so it is pretty similar :)

-Jeff

[Jesse Keating]

Post by Bill Perrotta
That is fine where can i download an iso of centos? If it is similar enough
to do all labs for rhel3 rhce that is my only concern.

Centos.org

[Matthew Miller]

Post by Jesse Keating
So why don't you use CentOS which as a annual or every other year release?

We use CentOS too. However, people a) want more cutting-edge and b) want
Fedora. And if my group doesn't provide something that covers that demand,
people will go off on their own, and then the security team will go back to
being hugely overloaded with Linux break-ins.

To expand on my earlier kvetching (sorry, no coffee yet):

I'm not able to force anyone here to do anything. Therefore, I have to
encourage good practice entirely via "carrots". This works best when we
align with the academic year -- a release in the spring, current through the
following summer to allow time for upgrades. Ideally, *two* years and a
summer, but I understand that's not practical.

As it is, what will happen is: whatever Fedora release is current as of
June-July-August will get installed on people's systems, and, with goading,
upgraded the next summer. If the actual Fedora release happens to be new in
June-July, the 13-month plan will be great, but if the latest release was
from, say, January, that leaves a big hole in which systems *will* get
broken into.

But, I find "If you need it to really work, use CentOS" to be a bad answer
for Fedora. CentOS is great, but since it is by necessity in its own world,
CentOS users don't feed back into the Fedora ecosystem in the same way,
which is a big loss for Fedora. (With the new baby, I missed out on
following the extras-for-RHEL discussion -- I need to check into how that's
panning out, and how it fits with merging Extras and Core. The availability
of Extras is currently a huge draw for Fedora over CentOS.)

So, given the realities, we probably will end up shifting our main BU Linux
efforts to Fedora, but may also provide a "BU Linux Extreme" Fedora spin.
I'm not sure how best to fit this into our calendar. If we disregard that,
we'll end up with insecure systems that just disregard *us*.

Extending the lifespan from ~9 to ~13 months is a huge help, but to cover
the gaps, we really need more like 18-19.

[Jeff Sheltren]

Post by Matthew Miller
Extending the lifespan from ~9 to ~13 months is a huge help, but to cover
the gaps, we really need more like 18-19.

If Fedora decides to officially support releases for ~13 months,
perhaps there is enough interest in extending them another 5-6 months
to keep Legacy going? If my thinking is correct, that would leave
legacy with 2 releases at a time, which *should* be manageable.. ;)

-Jeff

[Matthew Miller]

Post by Jeff Sheltren
If Fedora decides to officially support releases for ~13 months,
perhaps there is enough interest in extending them another 5-6 months
to keep Legacy going? If my thinking is correct, that would leave

Perhaps, yeah.

Post by Jeff Sheltren
legacy with 2 releases at a time, which *should* be manageable.. ;)

Another possibility would be to pick either even- or odd-numbered fedora
releases, and have Legacy only extend *one* of those.

[Jesse Keating]

If Fedora decides to officially support releases for ~13 months,  
perhaps there is enough interest in extending them another 5-6 months  
to keep Legacy going?  If my thinking is correct, that would leave  
legacy with 2 releases at a time, which *should* be manageable..

This is what we're trying to do now, and not succeeding very well. Or were
about to. I think there really needs to be significant interest in it, more
than just Matt Miller, although he is a very interesting case. The majority
of what we saw in the community was 'enough time to skip a release' and if we
can provide that, the need for more is no where near as much.

[Matthew Miller]

Post by Jesse Keating
about to. I think there really needs to be significant interest in it, more
than just Matt Miller, although he is a very interesting case. The majority

Yeah, because frankly, I have a _lot_ more interest than time. It's, like, a
10:1 ratio.

[Eric Rostetter]

Post by Matthew Miller
I'm not able to force anyone here to do anything. Therefore, I have to

That's the first problem... You either need to be able to force them
to do the right thing, or punish them for failure. If you can't do one
or the other of those then you're screwed, to put it bluntly.

Post by Matthew Miller
encourage good practice entirely via "carrots".

"sticks" work also. You get hacked, we unplug you from the network until
you comply. Gets their attention real fast when they are removed from
the network. Works better than carrots actually, in the long run.

Post by Matthew Miller
This works best when we
align with the academic year -- a release in the spring, current through the
following summer to allow time for upgrades. Ideally, *two* years and a
summer, but I understand that's not practical.

13 months for two versions gives you a lot of time IMHO...

Post by Matthew Miller
As it is, what will happen is: whatever Fedora release is current as of
June-July-August will get installed on people's systems, and, with goading,
upgraded the next summer.

Upgraded in the summer, whether the summer of the same or next year, and
the problem is solved for 99% of the cases... Only way that fails is
if you install early in the year (from January to April) and the next
release is done right after school (fall) starts that year... In which
case, that will hopefully be a small number of machines which you can
knock off during winter or spring break, leaving the majority until the
summer.

The draw back of the above of course is you need to track all the machines,
their versions, and installation dates, and keep that data updated. Basically
you need a good DB of the machine information...

Post by Matthew Miller
If the actual Fedora release happens to be new in
June-July, the 13-month plan will be great, but if the latest release was
from, say, January, that leaves a big hole in which systems *will* get
broken into.

If you install in January, then just upgrade in the summer if a new release
is out by then. See above for the rest of the details. I suppose there could
be a small hole, which is why most release cycles are 1.5 years instead of
1.08 years... But your call for 2.5 years seems way too long for a project
that wants to be cutting edge (and which you point out your users want
because it is cutting edge. If they want cutting edge, they need to upgrade
once a year, or else they are not cutting edge anymore).

Post by Matthew Miller
panning out, and how it fits with merging Extras and Core. The availability
of Extras is currently a huge draw for Fedora over CentOS.)

CentOS has Extras/Plus also for a lot of packages... And there are lots of
other packagers making "extras like" repositories out there... For the
"desktop" user this should be more than sufficient. It may of course
violate "server" or "production" users who have QA issues with that type
of thing.

In fact, one advantage of RHEL over FC/CentOS/anything-completely-opensource
is it actually comes with non-opensource software that is commonly desired,
and which is kept updated for security problems...

Post by Matthew Miller
Extending the lifespan from ~9 to ~13 months is a huge help, but to cover
the gaps, we really need more like 18-19.

I really disagree. The project is to be cutting edge, your users want
cutting edge, the only way to do that is to upgrade yearly. Otherwise,
both the project and your users are not cutting edge. If you can't
manage the upgrades in a year, then you need to hire more staff locally
(or better automate your upgrades).

Now, I really do feel for you and your situation. But I don't think you
can impose your bad situation on the Fedora Project, when you claim your
users really do want the same thing as Fedora Project, which means you
really do need to upgrade yearly, and not every 2.5 years. Fedora Legacy
is doing your users a disservice IMHO by not allowing them to be
cutting edge as they want to be.

In your case, I would think the only way to meet your needs would be with
Fedora Legacy, as Fedora Core just can't do 2.5 years of support and meet
its mission. But I'm not sure there are enough people in such a unique
situation, and who are so fixated on Fedora Core over other distributions,
to sustain something like Fedora Legacy.

Of course, I could be completely wrong... :)

[Matthew Miller]

Post by Eric Rostetter

Post by Matthew Miller
I'm not able to force anyone here to do anything. Therefore, I have to

That's the first problem... You either need to be able to force them
to do the right thing, or punish them for failure. If you can't do one
or the other of those then you're screwed, to put it bluntly.

Well, we're screwed, then. :)

Post by Eric Rostetter
"sticks" work also. You get hacked, we unplug you from the network until
you comply. Gets their attention real fast when they are removed from
the network. Works better than carrots actually, in the long run.

Oh, we do that if it comes to that. However, the goal is to avoid that in
the first place.

[...]

Post by Eric Rostetter
1.08 years... But your call for 2.5 years seems way too long for a project
that wants to be cutting edge (and which you point out your users want
because it is cutting edge. If they want cutting edge, they need to upgrade
once a year, or else they are not cutting edge anymore).

Well, as I said, 2.5 years would be ideal, but I recognize it to be not
really obtainable. I really would like, however, to see 1.5, or better, 1.6.

Post by Eric Rostetter

Post by Matthew Miller
panning out, and how it fits with merging Extras and Core. The availability
of Extras is currently a huge draw for Fedora over CentOS.)

CentOS has Extras/Plus also for a lot of packages... And there are lots of

Nothing like Fedora Extras, though. And third-party repos can be helpful but
coordinating them is work, and each requires a layer of maintenance of its
own.


[...]

Post by Eric Rostetter
I really disagree. The project is to be cutting edge, your users want
cutting edge, the only way to do that is to upgrade yearly. Otherwise,

Oh yes. In short, users want cake and they want to shoot themseves in the
foot with it.

Post by Eric Rostetter
both the project and your users are not cutting edge. If you can't
manage the upgrades in a year, then you need to hire more staff locally

Yes, it'd be great to be able to convince everyone I support to hire more
staff. That ain't going to happen.

Post by Eric Rostetter
(or better automate your upgrades).

There's significant engineer resistance to working towards making Fedora
yum-upgradable between releases. So that's really a non-starter.

Post by Eric Rostetter
Now, I really do feel for you and your situation. But I don't think you
can impose your bad situation on the Fedora Project, when you claim your

Clearly I'm in no position to impose anything. However, it'd certainly be
helpful to us if Legacy could contine to extend the lifespan beyond the new
proposed 13 months. And I mention it in case I'm not alone. [*]




* In fact, I'm pretty certain I'm not, and that there are thousands of users
running FC1, FC2, and FC3 and just waiting to become botnet members if
they're not already. The difference is that my users have me to care about
them.

[Jesse Keating]

Post by Matthew Miller
Clearly I'm in no position to impose anything. However, it'd certainly be
helpful to us if Legacy could contine to extend the lifespan beyond the new
proposed 13 months. And I mention it in case I'm not alone. [*]
* In fact, I'm pretty certain I'm not, and that there are thousands of
users running FC1, FC2, and FC3 and just waiting to become botnet members
if they're not already. The difference is that my users have me to care
about them.

But is there enough "you" to go around to do the updates? That's the real
question here. We can't stop people from being dumb and not upgrading their
release when it goes dead. If we gave them 2 years, they'd take it and still
not upgrade and then what? I think the Fedora project is really trying to
reach a reasonable amount of time that it can throw its entire support
behind. The Legacy folks of past and present are totally awesome, and we
could really use their knowledge and eagerness to help in perhaps more
productive fashions.

[Matthew Miller]

Post by Jesse Keating
But is there enough "you" to go around to do the updates? That's the real

Speaking for me personally, no. :)

Post by Jesse Keating
question here. We can't stop people from being dumb and not upgrading their
release when it goes dead. If we gave them 2 years, they'd take it and still
not upgrade and then what? I think the Fedora project is really trying to

Oh, that's for sure -- I routinely see incredibly ancient RHL machines still
in production. However, there's a curve, and the shorter the lifespan the
more people will be caught in it. At the current 9 months, I wouldn't be
surprised if it actually includes the majority of users. At 13 months, not
so bad -- but still a huge amount.

[Eric Rostetter]

Post by Matthew Miller
* In fact, I'm pretty certain I'm not, and that there are thousands of users
running FC1, FC2, and FC3 and just waiting to become botnet members if
they're not already. The difference is that my users have me to care about
them.

Well, I agree there are _at least_ thousands of users running FC1-3 waiting
to be botnet members. And most of them neither know about Fedora Legacy
nor have someone like you to care about them, so extending support via
FL won't help the majority of them... Sad, but no doubt true...

That doesn't mean FL shouldn't exist. Even if we can only stop a small
percentage from being hacked, that is still a very worthy goal which will
have positive effects on the internet/world.

However, if you can find a sufficient number of those who do know (or will
learn) about FL, or who have people like you who will care for them, and can
get them to support FL is some way, then there would be no problem keeping
FL alive to do so. But there has to be a certain level of support, and
I really don't see that happening myself. Again, I could be wrong...

The reason there was so much RHL 7/9 and FC 1 support was Red Hat really
dropped us with almost no advanced notice. We were aleady on the Red Hat
gravy train and the rug was yanked out from under us, and we had little
choice. I don't consider people installing Fedora Core 3/4/5/6 to be
in that same boat: they should know what they are getting into, and should
have a plan that meets their needs for the future. As such, there are not
so many people dependent on FC 3/4/5/6 and hence less people willing to do
the hard work for it.

Basically, FL was a _need_ for some of us at the start, but it isn't a need
for most people now. There are easy alternatives now that didn't exist
back when RHL was dropped and FC was started.

Anyway, I'm going to try to stop participating in this thread after this
message. I think I've said more than I should have already...

I'd be more than happy to see FL survive. I'd even be willing to help out
some. But it is no longer something I need, just something that gives
me a warm, fuzzy feeling...

[Nils Breunese (Lemonbit)]

Post by Matthew Miller
I'm not able to force anyone here to do anything. Therefore, I have to
encourage good practice entirely via "carrots". This works best when we
align with the academic year -- a release in the spring, current through the
following summer to allow time for upgrades. Ideally, *two* years and a
summer, but I understand that's not practical.
As it is, what will happen is: whatever Fedora release is current as of
June-July-August will get installed on people's systems, and, with goading,
upgraded the next summer. If the actual Fedora release happens to be new in
June-July, the 13-month plan will be great, but if the latest
release was
from, say, January, that leaves a big hole in which systems *will* get
broken into.

Every system needs an admin. I don't think it's realistic to not run
'yum update' for a year and expect everything to be fine. If you'd
run Windows on those systems you'd have to run Windows Update once in
a while. Now, I know Linux is less likely to get hacked very fast,
but like I said: every system needs an admin. If systems don't have a
proper admin, *then* they'll get hacked. This is not a Fedora-
specific issue.

Nils Breunese.

[Matthew Miller]

Post by Nils Breunese (Lemonbit)
Every system needs an admin. I don't think it's realistic to not run
'yum update' for a year and expect everything to be fine. If you'd

If there's no updates available, it doesn't matter how often they run yum
update.

[Nils Breunese (Lemonbit)]

Post by Matthew Miller

Post by Nils Breunese (Lemonbit)
Every system needs an admin. I don't think it's realistic to not run
'yum update' for a year and expect everything to be fine. If you'd

If there's no updates available, it doesn't matter how often they run yum
update.

That's why every system needs an admin (and not a nightly yum cron
job). A real admin will know or notice there are no updates available
and take appropriate action.

Nils Breunese.

[Matthew Miller]

Post by Nils Breunese (Lemonbit)
That's why every system needs an admin (and not a nightly yum cron
job). A real admin will know or notice there are no updates available
and take appropriate action.

Preaching to the choir. However, there's reality for you.

[Mike McCarty]

Post by Gene Heskett
I can't help but agree that its too short. 3 or 6 would be much more
realistic from the users viewpoint, who has his setup all fine tuned and
doesn't want to go thru that on an annual basis. There are other things
to life you know.

Yeah, like repairing vintage tube radios!

Mike

[Gene Heskett]

Post by Mike McCarty

Post by Gene Heskett
I can't help but agree that its too short. 3 or 6 would be much more
realistic from the users viewpoint, who has his setup all fine tuned
and doesn't want to go thru that on an annual basis. There are other
things to life you know.

Yeah, like repairing vintage tube radios!

Chuckle, how close to right was I?

Post by Mike McCarty
Mike

[Eric Rostetter]

Post by Matthew Miller

Post by Eric Rostetter
First I would like to say to those who say Fedora Legacy has failed, that
it _did_ work (i.e. didn't fail) for the most critical time period and the
most critical OS version (RHL 7-9, FC1). If it has failed, or is failing,
it must not be forgotten that before it failed it worked exceedingly well.

Or at least moderately well. Let's not over-sell. :)

I really think we did do incredibly well to start. We were often faster
than others, and often had less bugs than others (e.g. Progeny, spelling?).

We've only had two major problems with releases (one where a kernel install
failed to update lilo correctly but worked with grub, and one where sendmail
didn't work correctly on some older RHL upgrades). We did have to dump
RHL 8 quickly, and later RHL 7.2, but we stayed strong for a couple of
years on RHL 7.3 and RHL 9. And did a pretty darn good job on FC1 also
IMHO.

Now we don't have much demand for RHL any more, and we've failed pretty
bad to get any timely updates out for FC 2 through FC 4, but that isn't
the initial period I was talking about. That is the current situation,
which I admit hasn't gone well...

Post by Matthew Miller

Post by Eric Rostetter
Second, I'm fairly comfortable with saying that if FC goes to a 13 month
support cycle, FL is basically not needed anymore. IMHO, people can upgrade
once a year when presented with a known/documented release cycle, and known
documented alternatives.

One month of annual overlap is still a bit short.

_IF_ you want to use Fedora Core, you need to be willing to upgrade once
a year. And the 13 month window gives you just this amount of time to
upgrade. If you can't upgrade once a year, then you most certainly should
not be using Fedora Core.

My problem has always been I work in University settings where updates only
happen during breaks (Spring break, Summer break, or Winter break). On the
current Fedora Core schedule, a release can come at any time, and leave me
unprotected (if not for FL) until the next University break comes along.
With 13 months, I can easily stay with a release until the next break period
when I can upgrade. I really don't see a problem with the 13 month support
period, given Fedora Core's mission of being cutting edge. You can't be
cutting edge, free/community-based, and support a release more than a year
or so...

[Matthew Miller]

Post by Eric Rostetter
My problem has always been I work in University settings where updates only
happen during breaks (Spring break, Summer break, or Winter break). On the

Same here -- except I'm not sure I can rely on people to update during the
spring and especially winter breaks, or that the 13th month hits the summer
break in a convenient way.

[Eric Rostetter]

Post by Matthew Miller

Post by Eric Rostetter
My problem has always been I work in University settings where updates only
happen during breaks (Spring break, Summer break, or Winter break). On the

Same here -- except I'm not sure I can rely on people to update during the
spring and especially winter breaks, or that the 13th month hits the summer
break in a convenient way.

I don't have to "rely on people to update during the spring and especially
winter breaks" since I'm the only one who does the upgrades, and I can
usually depend on myself. ;)

Of course, this doesn't scale well... Fortunately I only have 80 machines
to worry about, so it is no big deal. If I had to do this in such a short
time period with say 200+ machines, then we'd have a problem...

[James Kosin]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Jesse Keating
See my blog regarding the future of Legacy as a project. Please remember
these are just proposals and not final solutions. A wiki page will follow
soon.
http://jkeating.livejournal.com/#entry_34659

I hope this doesn't mean Legacy is going away for good.

I may have some critical things to say about participation; but, I
still believe the community of participators can support a 6-12 month
window with the FC releases fall aside. This will give those who
choose to wait for a few months before upgrading the chance to keep
up-to-date with security fixes while they wait.

- -James
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFWxr+kNLDmnu1kSkRAuCoAJ94wq8mwcSaUorE92KkFk2QDqPDvgCfTxH4
pnYlEDi+uZoFFI97hPTkn8o=
=wnDn
-----END PGP SIGNATURE-----

[Jesse Keating]

Post by James Kosin
I may have some critical things to say about participation; but, I
still believe the community of participators can support a 6-12 month
window with the FC releases fall aside.  This will give those who
choose to wait for a few months before upgrading the chance to keep
up-to-date with security fixes while they wait.

The Fedora project would be offering 13~ months of updates (security only for
the last part), which gives you the opportunity to go from say Fedora 7 to
Fedora 9 + 1 month.

[James Kosin]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Jesse Keating
The Fedora project would be offering 13~ months of updates
(security only for the last part), which gives you the opportunity
to go from say Fedora 7 to Fedora 9 + 1 month.

I though Fedora project would be keeping only one previous version, so
by FC9, FC7 would have been taboo to even mention. And with the quick
pace of releases, 13 months would be about FC13 or 14 by then.

Just kidding....

Hmmm..... maybe a better upgrade path would be in order. Allowing
users to keep their configuration; with minor changes and upgrade the
units to FC6->FC7->FC8 etc. without any troubles.

I'll have to give that a try someday.

- -James
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFWyfxkNLDmnu1kSkRAvk7AJ0SotwaV8aSMNlS+kg1AsHqr8geigCePq4e
dVUQir8Q6T2Z6xyrSmvUhtk=
=Ixre
-----END PGP SIGNATURE-----

[Jesse Keating]

Hmmm.....   maybe a better upgrade path would be in order.  Allowing
users to keep their configuration; with minor changes and upgrade the
units to FC6->FC7->FC8 etc.  without any troubles.
I'll have to give that a try someday.

We're also committing to testing FC6 -> FC8 in one jump.

[Kirk Pickering]

Post by James Kosin
Hmmm..... maybe a better upgrade path would be in order. Allowing
users to keep their configuration; with minor changes and upgrade the
units to FC6->FC7->FC8 etc. without any troubles.

I am trying the procedure at the link below right now on a FC4
laptop. It's a method to upgrade from FC4 to FC5 via yum. So far,
it seems to be working well on a fairly clean FC4 installation.

One nice aspect of doing it this way is that I only have to
download 710MB of package updates, as opposed to 5 CD-ROMs
stuffed with packages that I don't need.

Has anyone on this list tried the following method?

http://www.makuchaku.info/blog/how-to-upgrade-from-fc4-to-fc5-via-yum

-Kirk

[Rahul Sundaram]

Post by Kirk Pickering

Post by James Kosin
Hmmm..... maybe a better upgrade path would be in order. Allowing
users to keep their configuration; with minor changes and upgrade the
units to FC6->FC7->FC8 etc. without any troubles.

I am trying the procedure at the link below right now on a FC4
laptop. It's a method to upgrade from FC4 to FC5 via yum. So far,
it seems to be working well on a fairly clean FC4 installation.
One nice aspect of doing it this way is that I only have to
download 710MB of package updates, as opposed to 5 CD-ROMs
stuffed with packages that I don't need.
Has anyone on this list tried the following method?
http://www.makuchaku.info/blog/how-to-upgrade-from-fc4-to-fc5-via-yum

The guy on the blog is my colleague. We do it all the time but it doesnt
work well with random packages and repositories. When Fedora Core
adopted the extras packaging guidelines a number of packages didnt have
a proper upgrade path and I had to do some post upgrade manual clean up.
It isnt yet good enough for me to recommend to new users.

Rahul

[Michal Jaegermann]

Post by Kirk Pickering
Has anyone on this list tried the following method?
http://www.makuchaku.info/blog/how-to-upgrade-from-fc4-to-fc5-via-yum

You can do that but how easy/straightforward that be depends very
much on what you got installed on a box and from where.

I did something of that sort bringing a box from FC3 to FC5, and I
finished with a working and correctly upgraded system; but this
required some careful thinking and planning of stages, and in some
moments the system was obviously broken but enough was working to
make the progress possible. With a "bad move" somewhere you may get
stuck. Not for a "general public" or faint in the heart.

Looks to me that this is in "if you have to read instructions how to
do that you better not try this at home" category. Sure, sometimes
it will work even if you will just follow instructions.

Michal

[Jeff Sheltren]

Post by Jesse Keating

Post by James Kosin
I may have some critical things to say about participation; but, I
still believe the community of participators can support a 6-12 month
window with the FC releases fall aside. This will give those who
choose to wait for a few months before upgrading the chance to keep
up-to-date with security fixes while they wait.

The Fedora project would be offering 13~ months of updates
(security only for
the last part), which gives you the opportunity to go from say
Fedora 7 to
Fedora 9 + 1 month.

I like this idea, and I'd be happy to see official support for an FC
release last ~13 months.

Of course, this would end all interest I have in Fedora Legacy, which
at this point is mostly to allow upgrading (well, re-installing in my
case) from FCN to FCN+2.

How much of this is just speculation at this point, and how close is
this to being actual policy?

- -Jeff

[Jesse Keating]

Post by Jeff Sheltren
I like this idea, and I'd be happy to see official support for an FC
release last ~13 months.
Of course, this would end all interest I have in Fedora Legacy, which
at this point is mostly to allow upgrading (well, re-installing in my
case) from FCN to FCN+2.

So by having 13~ months you would be able to do FCN to FCN+2

Post by Jeff Sheltren
How much of this is just speculation at this point, and how close is
this to being actual policy?

Depends on your feedback (:

[Matthew Miller]

Post by Jeff Sheltren
How much of this is just speculation at this point, and how close is
this to being actual policy?

Don't get me wrong -- this is definitely a positive development.

[Jeff Sheltren]

Post by Jesse Keating

Post by Jeff Sheltren
I like this idea, and I'd be happy to see official support for an FC
release last ~13 months.
Of course, this would end all interest I have in Fedora Legacy, which
at this point is mostly to allow upgrading (well, re-installing in my
case) from FCN to FCN+2.

So by having 13~ months you would be able to do FCN to FCN+2

Yes, that's exactly what I'm saying - that's what I want to do, and
having ~13 months of support would allow that to happen.

Post by Jesse Keating

Post by Jeff Sheltren
How much of this is just speculation at this point, and how close is
this to being actual policy?

I'm all for it.

-Jeff