Log in

No account? Create an account

Previous Entry | Next Entry

Fedora LTS?

There's been much discussion over the last few days over Fedora LTS because of a recent wikipedia conversion that took place.  I thought I'd give my 2 cents.

I'd honestly rather not have a legacy project though I'd never dream of preventing someone else from doing it.  Fedora isn't like the other OS's.  We move quick, we take risks.  And yes we alienate some users in the process but to me its worth it.  If we shift focus, even a bit, to be more mainstream, then our developers time get spent in other places which slows Fedora down.  Which, in turn, slows the industry down.  The kind of talent, time, and resources it takes to do something like Fedora legacy is absolutely huge and I'd rather see the people that have those skills remain focused on Fedora proper, the industry technology leader.

I think Fedora serves a very critical niche in the Linux ecosystem.  Without Fedora being the quick moving place it is, every other distribution would suffer for it.  So much stuff comes to Fedora first only to make it or die, it gets here first because we've positioned ourselves for it.  We can take the risks that the others can't and we do.

So then, what would I say to our users?  I'd say don't think of Fedora.  Think of the 'Red Hat family of products'.  If Fedora is moving too quick or not supported long enough for you take a look at Red Hat Enterprise Linux or CentOS.  No operating system can be everything for everyone and to try to be is foolish.

I also think that generally Fedora is much more about the process and the science we do and much less about the product we put out.  I use Fedora on all of my machines at home but if I had to choose between using Fedora and making Fedora, making Fedora would win out every time.  I suspect I'm not alone on that.


( 14 comments — Leave a comment )
Oct. 11th, 2008 09:10 pm (UTC)
You know it one of the bigger considerations over what path that fedora should take. I think that an LTS for fedora is a little too drastic a step for the distribution to take. But on the flips side, something in the how we are directing others in the community is off.
Maybe this is a marketing matter, since it more about how Fedora (and RHEL-clones) sell themselves to others. Because if something isn't clear in the wikipedia conversion, then I would suspect that something in the way that we point others to the clones that is being lost.
But you know what... It not as distressing an issue as people think.

I will admit another thing, there is something oddly unattractive with CentOS. I would guess it is the massive sense of isolation that comes with the Distro, in that for the most part you are on your own with any issue... that might be a lie, or not, it's hard to tell since CentOS is not nearly as vibrant as Fedora. So were I know what to do with issue in Fedora, CentOS is more of a mystery to me. But again... aside from pointing this out to CentOS, I don't think that there is anything that we can do about it. So, lets see how they react.
Oct. 11th, 2008 10:13 pm (UTC)
Re: Well....
IMHO, Fedora and CentOS need to join forces a bit more closely and promote one another. We are already doing that to some extent with things like EPEL (where it's very easy for Fedora packagers to build for RHEL and CentOS repos), but I think we can do more. The exiting part about Fedora to me is not Fedora 10 or latest or even the new desktop features but the community around the development of the software, much of that software also targets EPEL. It's just that we're nice and open and accomodating and have a lot of contributers -- the code to me is not as important. I would be in favor with having slower release schedules, actually, as the features I see moving forward in each release as long as folks continue to role out new software in updates. This in essense is what I liked about Debian (testing and/or unstable), you installed one release and essentially the software changed, but there weren't major updates that involved a "re-run installer event". This keeps people from running Fedora and the alternatives are too slow. I've heard the "Fedora moves too fast" argument from lots of folks who needed it on their desktops, and went the /other-distro/ route because the alternative moved too slow. So some middle of the road action would be advisable -- as what we are losing there is a potential contribute base. Compromises must be made in some places in order to cast a wider net; it will pay off later. The idea of an operating system that is only for developers seems to be near sighted. That all being said, I think we're mostly solid for users already, the only issue is the release cycle being a pain unless you're developing on those boxes. So there's that problem and it's also a bit of perception. But do I want us to be the fastest moving distro? Yes. Just keeping updates for a release turned on for an extra six months might be a start. I know I wouldn't mind pushing some software updates out to f7 still, I've got a script to do it -- and packagers who didn't wish to update non-security things could decide to stop doing updates. -- Michael DeHaan (posting anon because I can't get id.fedorapeople.org to work here)
Oct. 11th, 2008 10:42 pm (UTC)
Re: Well....
One of the problems is that some of the CentOS folks seem to be actively pissing on EPEL. I don't find that constructive. It is actually quite damaging if you consider how much it compliments their efforts.
Oct. 12th, 2008 03:54 am (UTC)
Re: Well....
Again, CentOS isn't the only clone working at the moment. I would also wonder about StartCom if possible.
Oct. 12th, 2008 01:54 pm (UTC)
Re: Well....
I completely disagree and do not get this feeling, and this comes from being around 100's of CentOS users on a regular basis.

There is much we gain from their presence than if a free-no-support-RHEL did not exist -- the community for our tools and software is much larger. They know RHEL also allows them to exist.

It's symbiosis.

-- mpd
Oct. 12th, 2008 03:52 am (UTC)
Re: Well....
Well CentOS does need to co-ordinate with Fedora a little more. Particularly with EPEL, if only to help use understand the need of CentOS in the scope EPEL. But then there is something else with EPEL that I wonder about... CentOS is not the only RHEL-based distro out there. I wonder if Scientific Linux, or White Box has any issues with EPEL (since EPEL covers Enterprise Linux 4 and Enterprise Linux 5) that CentOS users have experienced or anything difference.
Oct. 12th, 2008 01:56 pm (UTC)
Re: Well....
EPEL issues /do/ seem to be a bit silly when I run across them ... it seems to have a bit of a "NIH" view from some of the more "in" CentOS crowd, while the RHEL crowd loves it, and quite a good bit of the rest do too.

I hear silly things like "we should just use rpmforge" which is entirely bogus considering how badly dated some of the packages in rpmforge are.

So I think those are just people that don't quite understand yet. EPEL is awesome especially for the Fedora link.

Oct. 12th, 2008 03:24 am (UTC)
It simply wouldn't fit in with the Fedora way of doing things to keep a parallel track of legacy packages to be a Long Term Support model. There's RHEL to consider if LTS or 'support' is what is a main idea behind the argument. However, providing a means to the IHVs to pre-load Fedora might turn out to be an avenue worth exploring.
Oct. 12th, 2008 01:58 pm (UTC)
Don't consider LTS, consider instead keeping package updates turned on for 6 months longer.

That doesn't hold us back, that just allows for more options and expands the user base a respectable amount from those that are ok with upgrades but just don't want to do it so frequently.


Oct. 12th, 2008 05:18 am (UTC)
"No operating system can be everything for everyone and to try to be is foolish."

So, you think the universal operating system (Debian) is foolish?
Oct. 12th, 2008 07:13 am (UTC)
Debian is far from universal, common and universal are two different things. Debian is not the golden standard on what makes a Linux distribution... never has been, doubt that it ever will be with the different needs within the scope of computing. It has it's points, but even Debian has things it can improve on.
Oct. 12th, 2008 02:08 pm (UTC)
Re: Debian universal...
I've not heard of Debian universal but if it is trying to be everything for everyone then yes, I think it's foolish. To try to write something that is:

1) Stable enough for use in heart monitors in hospitals
2) Uses the latest software available
3) Supported for a long time 7+ years
4) Ideal for use on large machines with 128 cpus
5) Ideal for use in kindergarten classrooms

The bottom line is that in practice, many of the requirements listed above are in direct conflict with each other. It's just a matter of reality.
Oct. 13th, 2008 03:53 pm (UTC)
Well said!
Oct. 13th, 2008 07:01 pm (UTC)
LTS not really, Sane upgrades definitely
I use Fedora on all my home machines and in production on most machines in the workplace. Unless a system is very adverse to changes, the 13 (or so) month lifecycle of Fedora has always been fine for me.

Some systems which run more proprietary or finicky applications live on Solaris or RHEL. Some systems which have a reduced software footprint can (I think) live beyond the 13 month cycle safely.

The progress in Fedora is great, but if it begins to be considered "just a testbed" it's going to suffer. Let's please avoid that.

Keep it up. No new Fedora-Legacy. I'm happy as long as effort is put into sane upgrades, and that's always been the case.
( 14 comments — Leave a comment )

Latest Month

July 2013
Powered by LiveJournal.com
Designed by yoksel