[ale] kernel numbering

James Sumners james.sumners at gmail.com
Fri Mar 11 23:00:20 EST 2005


How am I supposed to know which kernel to use any more? The initial
system worked and Linus admitted as much in the email that was linked
in this thread. A kernel with an even point release number was stable
while kernels with an odd point release numer were not. True, the
releases were too slow under the old system but instead of keeping the
stable/unstable number system and adding in a definite time schedule
they just threw the old system out the window.

They (they being the ones responsible for the kernel tree maintenance)
replaced what worked with a system that no one can trust. They moved
to a system where every release of the kernel could include sweeping
changes that are likely to be bug ridden and completely incompatible
with the _minor_ revision before it. Now, they are admitting that this
is a bad scenario and trying to invent something else that is similar
to the old convention but still allows for much confusion. Hell, the
only thing I was able to figure out from Linus' email is that the old
system worked, the new one has people pissed off, and that they are
going to do something else now. I don't understand what it is they are
doing now but it sounds like a mixture of the old and new.

It is a confusing mess and is, in my humble opinion, a very bad mess.
What we are going to end up with is distribution specific kernels.
Companies like Red Hat and Novell are going to end up saying something
to the effect of "Well who knows what the hell the next kernel version
is going to contain so let's just branch from 2.6.whatever and
maintain our own version." That, in turn, will lead to third party
hardware developers supporting only specific distributions at the
driver level and leaving the vanilla kernel users to fend for
themselves. Personally, I was enjoying the fact that companies like
NVidia were starting to natively support Linux and that I could use
their drivers with a vanilla, stable, kernel. With the new "let's just
put whatever the hell patches in" mentality of the kernel I don't know
what to expect and neither do the people willing to develop for it
that are not part of the kernel maintenance team.

Like I said earlier, I don't trust the kernels past 2.6.7. 2.6.7 is
basically the last kernel version that was developed under the old
method. From what I understand, later kernels break support for
non-root cd burning and who knows what else. It used to be possible to
know if something major had been added to the stable kernel because
the backport would make the news in the release. Now, you have to sift
through the _massive_ changelog to determine what major things have
changed.

I just plain don't like it.


On Fri, 11 Mar 2005 22:35:00 -0500, Barry Hawkins <barry at alltc.com> wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Jonathan Rickman wrote:
> | On Fri, 11 Mar 2005 15:41:19 -0500, James Sumners
> | <james.sumners at gmail.com> wrote:
> |
> |>Basically, there is no sense in trying to keep up with the kernel
> |>numbering any more. I am still running 2.6.7 on my machine because
> |>there are not any remote exploits that I am aware of and I don't trust
> |>the later kernels. It used to be that I could upgrade to the newest
> |>kernels in the stable branch because they were just that -- stable.
> |>Now, who knows which kernel is stable and which is the testing because
> |>it is all in the same branch.
> |
> | And people wonder why I'm spending so much time testing Solaris 10...
> ~    Huh, I usually switch to a kernel on PowerPC after it has been out a
> few weeks with little to no trouble.  When you run on something besides
> x86 architecture, relying on older kernels means sacrificing performance
> and/or functionality for your hardware.  Plus, with the robustness of
> today's bootloaders it's not like you couldn't boot into your previous
> "sure thing" kernel if you have problems.
> ~    If folks want to escape the sometimes rocky road of ongoing kernel
> development, Mac OS X, Solaris, and even Redmond are always options.  8^)
> 
> Shrug,
> - --
> Barry Hawkins

-- 
James Sumners
http://james.roomfullofmirrors.com/

"All governments suffer a recurring problem: Power attracts
pathological personalities. It is not that power corrupts but that it
is magnetic to the corruptible. Such people have a tendency to become
drunk on violence, a condition to which they are quickly addicted."

Missionaria Protectiva, Text QIV (decto)
CH:D 59



More information about the Ale mailing list