Debian Release Mgr. Proposes Dropping Some Archs

smerdyakov writes "In this story posted by Andrew Orlowski of the Register Debian Release manager Steve Langasek has announced that support will be dropped for all but four computer architectures. Among the reasons cited for doing this are improving testing coordination, 'a more limber release process' and ultimately a ('hopefully') shorter release cyle. The main architectures to survive will be Intel x86, AMD64, PowerPC and IA-64." Actually, the story says clearly that this is only a proposal at this point, but it's definitely something to watch.

Comments

The hell?

[Lisandro]

Is it April the 1st already?

Re:The hell?

[temojen]

On April 1st NetBSD would be the one saying they'll only support 4 architectures.

Re:The hell?

[Anonymous Coward]

THANK THE LORD!

Someone at Debian is finally getting a fucking clue. I've been telling stupid Debian zealots this for years... your distro is dying because everything has to move in lockstep. Take a look at the Linux kernel -- it's x86, and yet there are loads of ports which move at their own speed. Debian is a slug of a distro because it moves at the speed of the absolutely *LEAST* developed port. Split them off focus on the x86 distro... and let the other catch up or die off. Debian is smothering... and all the puffed up insane zealotry about how other platforms are supported just as well as x86 is worthless if your distro is 5 years out of date.

The real headline..

[Anonymous Coward]

"Affected Admins Propose Dropping Debian"

Re:The real headline..

[Anonymous Coward]

It's like dropping acid, except you hallucinate that you're in an endless hallway, walking towards a light at the end called a 'major version release', while a vicious kid named Sid is whipping you.

Those would be the good ones to keep...

[thepotoo]

Seeing as they're the major systems out there. But IA-64? I've barely heard of that, and TFA says Microsoft dropped XP for that. Can anyone elaborate as to why this one was kept?

Re:Those would be the good ones to keep...

[Cramer]

I beg to differ. As a sparc/linux user and kernel hacker, the linux kernel is supported on sparc (sparc64 at least, sparc32 really is some dead-end hardware.) Granted, there aren't 10,000 developers maintaining it -- there doesn't need to be -- but it is maintained. The live development kernel (bitkeeper) has been usable for a very long time on sparc. So, either you aren't using sparc/linux, you're on sparc32 hardware, or you're just very unlucky. For the record, there are many x86 users that are frequently broken, too.

The lack of a SPARC maintainer is a concern, but one that can easily be addressed. (politics aside.)

Re:Those would be the good ones to keep...

[Anne Honime]

However, sparc far better supported in NetBSD and OpenBSD then it is in Linux.

Why use linux on sparc 32 in 3 letters : SMP.

Fine hardware, dead cheap, and NO bsd was up to it (until recently, if it happens to work now).

Debian back out in that area is a stab in the back for any user.

nooooooo

[Anonymous Coward]

so i won't have debian in my toaster????

well, I can still be using NetBSD. Of course the toaster runs it!

Dropping ARM???

[LWATCDR]

That might really hurt embedded developers. Seems like embedded users would be far more likely to use Deb than IA-64 users.

Re:Dropping ARM???

[pavon]

Actually, they didn't make a decision to cut or keep any particular distros, contrary to how the summary makes it sound. The actual email is worth reading.

The way I understood it was that each architecture would continue to be included in unstable, but when time came to release stable, the architectures that were not up to snuff would not be included in stable. In other words, they are not going to hold off on releasing stable for the architectures that are ready just because some other, less actively devloped ones are not. This seems fair. If someone wants their favorite architecture to be included in the next stable release then they can volenteer to get it up to stable quality, or commit themselves to maintaining it (security patches) after it is released. Otherwise they can just keep using the unstable. This is better than forcing everyone to use unstable, by holding debian back from releasing stable on a timely basis.

The second set of requirements (for SCC) also make sense. If you have less than 50 users, or cannot support the infrastructure needed make mirrors, there is no reason that all the ftpmasters should have to mirror a full branch of code for you - it is overkill. Those 50 people can get together set up their own apt-get repository for their binary packages.

There are several things that I did not like about this plan however, like the non-merit-based requirement, of requiring a machine to be purchasable new. If there are people that are willing to do the work, who cares if the machine is in production or not.

I also don't like the fact that there is no official option for the less active arch's to make stable releases uncoupled from the main stable timescale. Suppose that a minor arch, has enough support to do a stable release every 3 years compared to the x86's 18 month cycle. Choosing to target every other stable release won't work because while there is twice as much time between releases the bottleneck is the time between feature freeze and release, and that will still be determined by the x86 team's (faster) schedule. Furthermore, all the stable releases for all the architectures really should have the same package versions. This will save effort supporting the releases in the future (security patches etc), and keep user confusion down to a minimum. One possible solution would be if they kept the requirements listed, but did not require them to be met at the same time as the x86 branch - let the architectures enter stable when they are ready, with a time limit of say 2 or 3 release cycles of the x86 branch.

In general, requiring all the architectures to walk in lockstep is a real problem that debian needs to fix, but they should do so in a manner that allows the less active architectures to continue to have stable releases at their own pace, while not holding back the x86 line.

Well...

[Pflipp]

...there goes my handy Sparc server...

About time

[Anonymous Coward]

Thanks a lot! This was about time, or else we would never get a new stable release. Lets just hope thats it gets further then just beeing an proposal...

As anal as Debian is, this is kind of sad.

[theolein]

I mean, debian is the only distro that supports all the exotic architectures. If debian only supports the main architectures in futre, what then will the difference be between them and SuSE, Mandrake, Ubuntu and Gentoo for that matter?

Re:As anal as Debian is, this is kind of sad.

[misleb]

In my experience, apt is not the same when implemented on other dists. One the surface the commands are the same, but the integration isn't there. Neither are the normal debian repositories. Like you can't upgrade the whole system with apt-get dist-upgrade.

-matthew

In other news...

[Jon Abbott]

In other news, the NetBSD team announced that they have successfully ported NetBSD to the abacus...

Re:In other news...

[Thud457]

Aren't porting and actively supporting two different things?

Hooray "limber release process!"

[Look KG486]

Can't wait to get my hands on the new, stable 2.2 kernel!

Oh, wait...

Scary but beneficial

[caryw]

While Linux is well known for being exteremly cross-platform, 99.9% of installs will be on one of those four architectures. It would make sense to concentrate solely on those four rather than adding support for every Amiga and 68XXX setup out there. Especially now with Debian becoming a very strong player in the linux server community (now that RedHat is concentrating mainly on paid contracts and has allowed Fedora Core to become bulky and buggy.)

Besides, if you really want to run *nix on your Atari go download NetBSD.
- Cary
--Fairfax Underground: Where Fairfax County comes out to play

Not quite accurate ..

[abrotman]

Original email

They seem to imply it is a proposal to drop the actual releasing after sarge .. They will still have support for the other architectures, but seem to imply it must meet certain criteria to be considered for release.

IMHO: requiring a level of 98% is too high and only releasing if you can still buy is rediculous. Debian still mostly compiles for 386(on x86) and it's hard to buy a 386 these days.

IA64?

[bstadil]

Why not let HP and Intel carry the support banner for IA86.

The few machines sold hardly matters. HP 'claims" they will sppnd $3B on IA64 over next 5 years surely they can afford to pay for Linux on this dud of a processor.

Or better still pay the Debian guys

This is not final

[alfino]

As an active Debian developer, I simply want to state: this is anything but final and not at all decided. I am only one of many developers against the proposed scheme, and especially against the way in which the scheme was devised -- in a closed meeting with only a few select members, and completely without soliciting any input from the community.

In the long run, Debian may well have to concentrate more on some architectures than others, but a radical step such as the one proposed will probably not fly well with the community. Since our users are our top priority, you can expect many more emails on the topic before anything will happen.

Re:This is not final

[Ulric]

I'm not a Debian user, but I understand that the long release cycles are viewed as a problem by those who are. Do you think it will be possible to solve that problem without dropping a few architectures, and if so, how?

I understand that Gentoo supports several architectures, including several (alpha, sparc) that would not be supported with this scheme. How come they don't seem to have a problem getting releases out the door? (You may not have more of a clue than I do, but perhaps someone else does.)

Don't we already have that?

[Anonymous Coward]

It's called Ubuntu.

So what thing might mean

[El_Muerte_TDS]

is that only those 4 archs will be actively supported in Debian _releases_. Other architectures will still exist and maintained but not be included in the shifts from unstable->testing->stable.

If it's that it might be a good things, granting the more popular(?) architectures a smaller turnaround time for stable releases.

Or maybe hell freezes over.

What about ARM ?

[EmbeddedJanitor]

By the end of this year, the majority of Linux systems will be cell phones and settop boxes/ digital TV etc running on ARM and PowerPC architectures .... not x86. I would have thought that keeping ARM would be a GoodThing.

Perhaps Debian isn't trying to address the embedded segment.

IA-64 vs AMD64

[cbr2702]

Not at all. The IA-64 is Intel's Itanium architecture which was massively redesigned. It is not compatable at all with x86 or AMD64 and is actually closer to the PowerPC, as both are RISC chips. The Itanium hasn't done very well (IBM just stopped selling it for their own POWER arch) but it it still used, and probably is at least #4 on servers.

Damn.

[gt_swagger]

So much for running Linux on Bubba the Big Mouth Bass. That was my dedicated firewall too!

Re:NetBSD, here I come

[glen604]

gentoo runs quite a few mips architectures.. check out their support here:
http://www.gentoo.org/doc/en/mips-requirements.xml

Excuse me?

[Nasarius]

Gentoo supports at least as many architectures as Debian. A cursory glance at packages.gentoo.org will tell you that.

I disagree wholeheartedly.

[hummassa]

Someone at Debian is finally getting a fucking clue. I've been telling stupid Debian zealots this for years... your distro is dying because everything has to move in lockstep.
Interesting, from where I am it seems to be pretty much alive, thank you.
Take a look at the Linux kernel -- it's x86, and yet there are loads of ports which move at their own speed. Debian is a slug of a distro because it moves at the speed of the absolutely *LEAST* developed port.
There is always sid.
Split them off focus on the x86 distro... and let the other catch up or die off.
And then the only thing that sets Debian apart from the other distros (quality, determined by lots of portability issues spotted, bad code spotted this way, lots of different archs using the same distro, etc. will be dead. People will just use Ubuntu, if they want to use something x86-ppc only.
Debian is smothering... and all the puffed up insane zealotry about how other platforms are supported just as well as x86 is worthless if your distro is 5 years out of date.
Interesting, I run Debian, with kde 3.4 over kernel 2.6.10 and my distro does not feel 5 years out of date.

Re:I disagree wholeheartedly.

[noahm]

Interesting, I run Debian, with kde 3.4 over kernel 2.6.10 and my distro does not feel 5 years out of date.

Sure, you run sid. You know what that means? It means that this proposal won't affect you at all. (additionally, I'm sure you run x86, along with what, 98% of all other debian users?)

The thing is, you're the type of user who doesn't need predictable release cycles. You can get by on the bleeding edge and run software for which a new package release may be uploaded on any given day.

A lot of Debian users are in very different positions. I, for example, run Debian in an enterprise environment, with literally hundreds of servers and workstations. woody is simply not an option in this environment. Hardware support (both kernel and user space) is dreadfully lacking, and we'd have to backport most of the software we use every day anyway. We'd end up running something so bastardized that we'd no longer see many of the benefits of running Debian at all. So we were forced to go with something more current. We chose sarge, with the understanding that we'd have to be responsible for the security of our systems, with little help from Debian. But of course, there are problems there, too. Sarge changes every day. A machine installed today may look nothing like a machine installed tomorrow. Additionally, we simply have no way of knowing when sarge will be released. The saying within Debian has always been "we'll release when it's ready", but of course, there's never a published metric for readiness, so there's simply no way of knowing when that will be.

Basically, right now, Debian really doesn't have a good release for enterprise users. That really sucks, since IMHO Debian provides a software infrastructure that makes it really appealing for large scale deployments. I really hope this new proposal is a step toward a shorter and more predictable release cycle!

noah

(Debian developer, sysadmin, and user since 1997)

Re:I disagree wholeheartedly.

[Kent Recal]

A machine installed today may look nothing like a machine installed tomorrow.

You may want to take a look at FAI (Fully Automatic Installation - google will find it). We've been using it quite successfully for that kind of maintenance.

You basically set up a local debian mirror (snapshot of the real tree) and use it to deploy your machines (FAI does it great) and as only apt-source for them. Whenever it's time to update a pkg you test it, then just drop it to your mirror where the clients can pick it up via apt-get upgrade.

Why keep IA-64?

[James Youngman]

Why bother keeping IA-64? Debian has more alpha users than ia64. There are more SPARC users. Heck, there are even more HPPA users than ia64 users. All the details are available at the Debian Popularity Contest.

misleading

[macshit]

The phrase "dropping support" is misleading. They're dropping the "stable" release for these archs. They are moved into a category called "second class citizen" architectures.

"unstable" -- which is what hacker-type individuals tend to run anyway (and is both much more up-to-date and not particularly unstable) -- will continue for all. As most of the affected archs fall into the "mostly for hackers" category, this change should have little real impact. I suppose the exception might be the sparc.

The benefit of all this is (besides, maybe, faster releases) that they have a plan for adding new scc archs easily.

[I think the "scc" archs will also not use the Debian mirror network, but probably don't have enough users to receive any real benefit from it either.]

Re:Why can't the kernel be seperated from the dist

[Just Some Guy]

My understanding is that it isn't so much the kernel (although that's certainly an issue) as the userspace applications. For example, going from 32 to 64 bits breaks a lot of badly-written software, as does that annoyingly still-present issue of endianness. Debian currently treats all platforms as equal, meaning that a problem compiling X.org on some weird 48-bit middle-endian system used by 15 people can delay including that package on x86 and x64 as well.

If everything was well-written and accounted for differing word lengths, byte orders, etc. then we wouldn't be having this conversation. Unfortunately, that's not the case. On the plus side, Debian's dedication to platform equality means that a lot of bugs get exposed (and fixed) that no one would ever know about if the world only ran x86. This is a good thing for everyone, even those where that software already worked as expected.

Re:Now...

[Qzukk]

Even the trolls can't be bothered to read the damned thing. Debian will still be available on all those plaforms, but Debian Stable won't be after Sarge releases.

If this proposal passes.