Re: Kernel Geeks Unite?

James Mastros (
Wed, 10 Sep 1997 00:06:29 -0400 (EDT)

On Tue, 9 Sep 1997, Jon M. Taylor wrote:

> On Mon, 8 Sep 1997, Robey Pointer wrote:
> > Rob Hagopian wrote:
> > >
> > > Should someone really have to upgrade to 2.1.x to use the ThunderLan
> > > driver [pof: the tlan driver hasn't been backported to 2.0.x yet]? I
> > > don't think so. Should someone really have to upgrade to 2.1.x to get
> > > better SMP performance and IPv6? I do think so.
> >
> > Stability and new features are mutually exclusive goals.
> Features and drivers are not the same thing in this context. I
> still get Win31 drivers with new printers even today. Making a general
> practice out of backporting drivers to the previous stable kernel release
> would go a LONG way towards keeping people from *having* to use
> potentially unstable development kernels just to get access to a needed
> driver.

This is because drivers in the Windoze world and the Linux world are handled
completely differently. In Linux, the drivers are part of the kernel;
ocasionally drivers are in sepperate packages, but these are considered
interum versions untill the newest version is in the kernel.

In Windoze, OTOH drivers are the responsibility of the divice maker. Some
drivers are included with the OS, solely for the ease of the user.

It would be wonderful if device makers would include linux drivers, but it
ain't gonna happen anytime soon. I propose we take a clue from the Windoze
world (sickening though it seems <G>) and make drivers sepperate modules,
with sepperate tarballs. All the kernel support we need is to make it
easyer to link a sepperate module into the kernel. How hard would it be to
make it possible to link a module into the kernel? (Assuming that the
module is compatible with that kernel version...

-=- James Mastros

> Jon

I can now be reached again at or
  "Shooting as [a] communications method is obsolete even here in Bosnia, so
I'll skip over it."
	-=- Dragisha Durich