Linux 3.0 Kernel May Remove Some Old Cruft

[stan]

With all due respects to Alan Cox, it's not the first time he's pushed for killing off useful drivers to make his life easier (remember libata vs ide debate?). Those are WORKING drivers that aren't hurting anyone, why does he want to artificially remove support? While Alan may be a great hacker, he also has this lazy/destructive attitude (towards things he personally doesn't care about) that will hopefully be kept in check by more rational kernel devs.

[pvtcupcakes]

Who Cares? I really don't get why people are so wrapped up in version numbers these days...

I'm not getting wrapped up in version numbers. I think that 3.0 should be just like any other 2.6 kernel release and that they shouldn't do anything special to make 3.0 seem like a major release.

I don't even know of anyone who would say that the 2.6 development model was a failure, so why would they want to change that now? If he wants to remove old drivers then Alan can wait until 3.1 or 3.2, there is no reason to hastily rip out these drivers when the merge window closes in less than a week. I think Linus would see it the same way.

I just went back to read the thread on LKML and H. Peter Anvin said it better than I could:
https://lkml.org/lkml/2011/5/24/398

And then Linus agreed.
https://lkml.org/lkml/2011/5/24/400

Indeed.

It's not about features. It hasn't been about features for forever.

So a renumbering would be purely about dropping the numbers to
something smaller and more easily recognized. The ABI wouldn't change.
The API wouldn't change. There wouldn't be any big "because we finally
did xyz".

Linus

[rapsure]

Modern motherboards use an ISA bus for small devices like sensors even though they do not have a physical slot.

http://en.wikipedia.org/wiki/Industr...embedded_chips

I already knew that because I've developed for embedded devices and sensors. The i2c bus is on the isa bus along with many other devices such as the old ps2 keyboard and mice controllers.

In the kernel config AMD64 just doesn't provide the option of selecting the isa bus but it is there. Many may not realize that but it isn't very important because the isa bus doesn't require a large kernel footprint and simple devices are connected to the isa bus.

I'll be fine with removing EISA, VLB, MCA, but removing ISA is just impractical. The EISA, VLB, and MCA buses are primarily found on 486 computers and were short lived. I suppose it is a question of how many 486 computers are running a modern linux distribution?

[schmidtbag]

jumping from a MAJOR version number (as in 2.6 all the way to 3.0) is deserves some notability. its one thing to say "o ya we're going to add nvidia optimus and move the version up to 2.8" but a few small things like that don't deserve an ENTIRE new version.

however, cleaning up the kernel with old stuff is a GREAT idea. aside from the fact that almost nobody uses that old stuff anymore, the linux kernel today is modern enough that it probably won't work well on something too old anyway. get rid of the old drivers, remove obsolete features, and maybe even find more efficient ways of doing something.

in programming, you can have 2 scripts perform the exact same task without any wasted, repetitive, or redundant code. one script could be 50 lines, but the other could exceed 100. i'm sure the linux kernel has many situations like this right now.

do this "spring cleaning" and i'd say the kernel deserves a step up to version 3.0

[Geri]

Yeah, remove isa bus, and loose support for almost all temperature / rpm / voltage sensor. Incompetence of the kernel developers triumphes again over the normal humans.

[DeepDayze]

I already knew that because I've developed for embedded devices and sensors. The i2c bus is on the isa bus along with many other devices such as the old ps2 keyboard and mice controllers.

In the kernel config AMD64 just doesn't provide the option of selecting the isa bus but it is there. Many may not realize that but it isn't very important because the isa bus doesn't require a large kernel footprint and simple devices are connected to the isa bus.

I'll be fine with removing EISA, VLB, MCA, but removing ISA is just impractical. The EISA, VLB, and MCA buses are primarily found on 486 computers and were short lived. I suppose it is a question of how many 486 computers are running a modern linux distribution?

I'm sure the ISA driver code will be kept for exactly that reason and I can see letting the EISA, VLB and MCA code go. Why not also dump those drivers for the old proprietary CDROM drives such as the old non-ide Panasonic, Mitsumi and Sony drives as these drives are hardly existent anymore.

[leif81]

If Linus wants to call 2.6.40 version 3.0 then that's good enough. 3.0 doesn't have to be special.

If it's simply rebranding then what's the point.

"3.0" will draw lots of media attention,etc so it hopefully will be more noteworthy than "it's just 2.6.40 but we named it 3.0". Use it as a chance to drop regressions, drop old drivers, use it as a chance to rearchitect things and generate interest.

[ammo42]

I'm sure the ISA driver code will be kept for exactly that reason and I can see letting the EISA, VLB and MCA code go. Why not also dump those drivers for the old proprietary CDROM drives such as the old non-ide Panasonic, Mitsumi and Sony drives as these drives are hardly existent anymore.

These drivers have already been deleted in Linux 2.6.23.
http://git.kernel.org/?p=linux/kerne...be49bfbd9e730d

[not.sure]

If it's simply rebranding then what's the point.

"3.0" will draw lots of media attention,etc so it hopefully will be more noteworthy than "it's just 2.6.40 but we named it 3.0". Use it as a chance to drop regressions, drop old drivers, use it as a chance to rearchitect things and generate interest.

No non-geek is interested in the kernel. People are interested in distributions.

The point of the rebranding is to get smaller numbers, plain and simple. It's about ergonomics and perception. Humans also prefer measuring time in years/months/days/hours rather than epoch time; very similar thing.

[leif81]

No non-geek is interested in the kernel. People are interested in distributions.

Firstly, how many non-geek linux users do you know? I know very few.

You can bet your ass every printed paper Linux mage will have a cover story if the kernel is renamed 3.0. And every linux distro that carries Linux 3.0 will list it as a headline "feature". Version numbers garner attention.

Aso, see Gnome 3.0 for how much a community can be rejuvinated for a big release.

Simply renumbering for the sake of renumbering is a waste of a MASSIVE opportunity.

P.S. Just mentioning to some of my non-linux friends that the new kernel may be named 3.0 got them all interested. And then I said nothing is new except the number. And so they went back to hacking away on Windows.