Re: [ANNOUNCE] Vendor kernels unpakced

[Marco Colombo <marco@esi.it>]

On Sat, 11 Aug 2001, Joseph Andrew Knapka wrote:

> Marco Colombo wrote:
> >
> > [removed l-k since it's OT there]
> >
> > On Fri, 10 Aug 2001, Phil Kos wrote:
> >
> > > > Just as it makes no sense to run a not-up-to-date release
> > > > kernel, it makes
> > > > no sense to keep anything but the very last on line. Just MVHO.
> > >
> > > I beg to differ, Herr Doktor von Brand. It makes more sense to run a
> > > not-up-to-date release kernel than it does to slavishly update to every new
> > > release without any pressing need for new features or functionality. Don't
> > > fall into the commercial software trap of "it's newer, so it *must* be
> > > better!"--here be tygers...
> >
> > What? Most updates are bug fixes.
>
> Except that many are enhancements to support future functionality,
> or refectoring of working code in order to improve maintainability,
> or complete rewrites of major, working kernel subsystems (eg VM
> changes 2.2-->2.4). Frequently the bugs that are fixed in a minor
> release are ones that are only triggered by uncommon kernel or
> hardware configurations. While quality control of the Linux kernel
> is generally quite good, it's not unheard of for such improvements
> to break the kernel for some subset of users.
>
> > And I *do* want kernel bug fixes.
>
> Even if those bugfixes have no relevance to your particular
> site? If the bugs that are fixed aren't relevant
> to your installation, the risk surely outweighs the benefit.
> If the machine behaves identically after the upgrade, what
> have you gained?

Since I don't have either the time or the knowledge to mantain the
linux kernel myself, even only the parts I need, I have to trust
someone else. I trust Alan in that 2.2.20 is a better kernel than 2.2.19.
I trust Red Hat in that 2.2.19-6.2.7 (or whatever) is better than
2.2.19-6.2.1.  Systems are simply too complex for me to believe I
can make a better job. I can't be sure that starting from a Red Hat 6.2
out-of-the-cd and performing only selective upgrades (including kernels)
(based on what *I* think I need) I can get a system that is more stable
than a 6.2 with *all* updates from Red Hat.

> Of course if your goal is not to maintain a stable production
> box, but rather to keep up a bleeding-edge development environment -
> also a perfectly reasonable goal - then go right ahead :-)

updates != new releases

> Security fixes are also a good reason by themselves to
> upgrade a kernel, although security problems seem more
> frequently to arise in user-space programs.

True, but that's not a reason for leaving holes in the kernel.

>
> Regards,
>

Gi

.TM.
-- 
      ____/  ____/   /
     /      /       /			Marco Colombo
    ___/  ___  /   /		      Technical Manager
   /          /   /			 ESI s.r.l.
 _____/ _____/  _/		       Colombo@ESI.it

-
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
IRC Channel:   irc.openprojects.net / #kernelnewbies
Web Page:      http://www.kernelnewbies.org/