Filesystems

Gabriel Gunderson gabe at gundy.org
Fri Sep 2 08:54:00 MDT 2005


On Fri, 2005-09-02 at 01:03 -0600, Shane Hathaway wrote: 
> Gabriel Gunderson wrote:
> > On Thu, 2005-09-01 at 22:50 -0600, Shane Hathaway wrote:
> > 
> >>(Tangent: I find Linus' kernels to be much more stable than
> >>distribution kernels; anyone else have the same experience?  This is
> >>contrary to the wisdom I hear on the net.  Maybe it's because I
> >>configure the kernel for specific machines rather than throw in every
> >>driver available.)
> > 
> > 
> > I think you will see less of this as time goes by.  I hear a lot of talk
> > about how they are not trying to deliver a finished product like you
> > would expect someone like Red Hat to deliver.  They realize that the
> > majority of kernels in the wild were packaged by a distro and they don't
> > seem to want to duplicate that effort when they could be working on the
> > next big thing.  I'm not saying you can't run a vanilla kernel, I'm just
> > saying that they are not trying to productize it.
> 
> Isn't the new four-level versioning system evidence of moving *toward* 
> stability in the vanilla kernel?  They saw a lot of distributions 
> duplicating the work of stabilizing the vanilla kernel, so they 
> centralized much of that work.

Andrew Morton on the subject...
"So, you can look at it that, I and Linus, we don't really make a
product.  We make a technology component - which is a *little rough
around the edges* but it is evolving very, very, very, very fast and the
productization of that technology is the service which the distributors
provide.  They will take our rough technology, stabilize it, ship it as
a product.  Um, it wasn't always the case. You go back a few years ago
and we, the public kernel developers, we divided quite a lot of our time
to - also to trying to ship an end product - that's this 2.2 kernel
thing - the 2.4 kernel thing.  We were always spending our time between
developing the raw technology and also trying to ship a product which is
further productized by the distros. Now we have sort of moved a lot of
resources away from that productization process nowadays.  Very much...
and most of our resources are concentrated on developing that technology
input."

> > Now, kernel stability should be up to your distro.  If you can't get a
> > stable kernel from them... then it's time to jump ship.  In three years
> > with Red Hat (and now CentOS) kernels I've only seen one (1) kernel
> > panic and it was my fault for messing with the modules for digium
> > hardware (zaptel).  I don't know how others have done with them. YMMV.
> 
> It was on Red Hat that I learned about kernel instability. ;-)

Like I said, your mileage may vary.

Gabe

> Shane
> .-----------------------------------.
> | This has been a P.L.U.G. mailing. |
> |      Don't Fear the Penguin.      |
> |  IRC: #utah at irc.freenode.net   |
> `-----------------------------------'




More information about the PLUG mailing list