Fedora Legacy Project, Thoughts on Shutdown

« previous entry | next entry »
Jan. 3rd, 2007 | 09:37 am

Let me present some credentials, I've been a longtime Redhat user. I switched to using Redhat somewhere around the 6.0 releases, before that I used Slackware. Sitting around for my use is OSX, one Solaris machine, and one box running Trixbox (which is CentOS underneath).

Everything that I do in a production fashion is running Fedora Core.

Have I had problems? Not really. I keep fairly current so I've not needed the Legacy Project. In the one case I do need it, for an RH 8.0 box, I consider the existence of the 8.0 box to be a problem (one I am actively trying to solve).

Why do I need it? Because I picked some software to use on the box that won't run on a modern FC setup. I bet on the wrong horse and I only stick with it because its a lot of effort to get off that horse. I'm working on it and I hope by the end of the month I can kill the 8.0 box, but I am not holding my breath.

While I don't need the Legacy project I know that a lot of others do. I use FC in large part because I see it as a way of being on a stable release but at the same time letting me stick my neck into the edge of open source development. Why do I want this?

I see it as a competitive advantage. I get the latest, I work with the latest, yet my stuff still works.

Years ago I use to watch Jeff and Rob run Debian on their laptops. It was a "so X work for you today?" sort of environment. I don't need this, and I don't want it.

A computer to me is about solving problems, not creating them.

The thing is every so often you bet on the wrong horse. You pick an application that doesn't upgrade. When that happens you need to be able to support the platform till you get off of it. If you are lucky this never happens, but you want that option. If you are especially smart you realize when this happens and toss out the piece which is aggravating the problem.

BTW I think MySQL Community is different for one big reason, we aren't a distribution. I listen to people compare us to Redhat, and I think they don't get it. We are software creators, not distributors. It works differently in our case. In our case Community gives us the ability to create something which is bleeding edge, something that is updated more often with features. Its a moving target and I find a moving target to be a good thing.

Why? Because the database is one place where I want every advantage I can get my hands on. The enterprise customers don't want this, they want something that is completely stable. Those who are doing web development and building software real time? Different case entirely.

Its the nature of the beast.

Yes, as Peter pointed out we have completely sucked at getting changes into the community tree. We are sitting on some great patches by Jeremy Cole that need to be merged into the community tree. I went in and merged up the 5.0 and 5.1 trees this morning so they are at least up to date with their respective upstream providers. The good news is that we have real intentions, the bad news is that we are being slow about it. One of our New Year resolutions needs to be to get better at this.

Link | Leave a comment | Share

Comments {5}

Ted Tso

(no subject)

from: tytso
date: Jan. 3rd, 2007 11:57 pm (UTC)
Link

I've moved from Debian to Ubuntu for the same reason; I get a userspace which is reasonably bleeding edge (releases every six months!) but yet everything Just Works.

I still am a Debian Developer, because even though I have regression test suites for e2fsprogs, there still are bugs that only show up when a larger community of users try it out for the first time, and the folks who are willing to use Debian Unstable make for a great set of guinea pigs.... so I will generally push out my changes to Debian unstable and let them bake for a few weeks before I cut a formal e2fsprogs release.

Reply | Thread

Brian "Krow" Aker

(no subject)

from: krow
date: Jan. 4th, 2007 11:25 pm (UTC)
Link

I am still holding off on using Ubuntu for servers but it looks like that will be the way I am going to go. I am tempted to bundle up my own rPath distribution, but I haven't decided yet if that makes sense (and I don't understand the rPath business model, which I need to...)

Now... I've never thought about this before but I wonder if it would make sense to shove a mysql-5.1 into Debian unstable to get a bit more testing...

Reply | Parent | Thread

(no subject)

from: jamesd
date: Jan. 6th, 2007 02:19 am (UTC)
Link

Yes, please do push 5.1 into Debian unstable. It needs all the testing help it can get before we say GA instead of after. :)

Reply | Parent | Thread

Ted Tso

(no subject)

from: tytso
date: Jan. 6th, 2007 03:18 pm (UTC)
Link

I've found that Ubuntu Edgy is actually quite nice as a laptop/client OS. I put it on Stacey's laptop, and she's ecstatic that with NetworkManager, you can actually click and select which Wireless network to use. No more opening a gnome terminal, manually using iwlist to see what networks are around, and then manually selecting one using iwconfig and then manually starting dhclient! (Never mind that MacOS has had this for years....)

Suspend/resume just works, as done 3-D acceleration and wireless cards that require binary firmware files. And the next version of Ubuntu is supposed to be focusing on multimedia functionality (although everything I've wanted to work has, so I'm not sure what they will be improving upon).

As I understand rPath's business model, their server for distributing their new packaging technology (which is supposed to make upgrading of an existing software application product painless) is proprietary. If you are distributing your packages for free, you can use their rBuilder online for free, but everyone gets to download your stuff. If you want restricted downloads, you have to pay them $$$. I think they are presuming that even if a company is selling support for OSS, using a Cygnus-type model, rPath is assuming that their customers aren't going to want to make security and errata upgrades completely painless to download; the OSS support-based company will want to restrict the automatic binary update services to paying customers, much like Red Hat does with RHN. (That's one thing I don't understand about Canonical's business model; given that you get security updates for free, will they get enough customers willing to pay $$$ for them to have a sustainable business model?)

Reply | Parent | Thread

Brian "Krow" Aker

(no subject)

from: krow
date: Jan. 8th, 2007 05:06 pm (UTC)
Link

I got this from Elliot Murphy of Conical:
Ted is right, security updates (and the versions of ubuntu meant for
servers) are free, and they will stay free. This is a powerful
differentiation from the other linux vendors. The business model has us
generating revenue from a number of sources, including 24x7 professional
support, derivations, and other value-added services. For example, we
recently did some interesting work with SFO airport where Ubuntu is
running the kiosks. I'm personally working on providing some additional
commercial services related to launchpad.net in the near future.

Reply | Parent | Thread