Mike Barnes wrote:
> <excellent reasoning snipped>
> So please don't think I'm griping and trying to get something for
> nothing. I'm not. All I want is to be able to use something I paid for
> no more than six months ago. I know we live in a world of fast upgrade
> cycles, but this is a bit too much.
The underlying problem appears to be the packaging of the Win4Lin kernel mods.
A minimal set of kernel hooks needs to be created in a way that Alan Cox and
Linus will find them suitable for inclusion in the base kernel (that is, either
the hooks need to be useful to a wider audience than just NeTraverse and
Win4Lin, or Win4Lin needs to become insanely successful, to the point that a
significant number of Linux users also use Win4Lin). Then the rest of
Win4Lin/Merge can become a kernel module.
Of course, it should be possible to rewrite Merge just enough to make it use
existing hooks, much like VmWare does. That may add a performance hit, but it
would allow those users needing the flexibility to detach kernel upgrades from
the Win4Lin upgrade cycle.
Either way, it seems clear that Win4Lin still has some crucial architectural
evolution yet to undergo, and until that's done, all the releases so far should
really be viewed as 1.x releases. The *real* 2.x release (no matter if
NeTraverse calls it 8.0 or 3.907) will work with any and all kernels, out of the
box.
Though I know nothing about VmWare, I suspect it may be a better choice for
people like Mr. Barnes. And for me (I've been having *hard* lockups with
Win4Lin since the Beta5 release of 2.0).
But only for the moment, I hope...
Of course, NeTraverse could simply create their own distribution and (mildly,
minimally) fork the kernel. Then they could get out of the
patch-everyone's-kernel business. If enough people switch to the NeTraverse
distro, well, that could encourage other distro vendors to include Merge
support, and would sure send a clear message to Alan and Linus.
It's all a matter of determination, and an awareness of the target audience. So
there are two paths that can be followed in parallel:
First, make a truly minimal and elegant patch (re-architecting Win4Lin as much
as needed), and submit it as often as needed until all arguments against it have
been properly resolved. And encourage all distro vendors to include the patch
in *all* their kernels, at least until it becomes part of an official released
kernel.
Second, create a distro with the patch included, and include the distro with
every shipment of Win4Lin (with install options to NOT use the full distro, of
course!). This will allow NeTraverse to decouple from other distros, and
provide a simplified path for those Win4Lin users willing to follow it. I'd
probably go for this path (for several reasons, the main one being that I run
Win4Lin on a spare headless system), and I suspect a significant percentage of
existing Win4Lin users would certainly consider it.
Both goals could be met by partnering with an existing distro vendor (such as
RedHat). Once Win4Lin-capable kernels become prevalent, and a major distro
vendor puts their weight behind it, other distro vendors may follow suit,
independent of what Alan and Linus decide. Getting other distro vendors on
board should also help the evolution of the patch itself, since nobody seems to
want a "real" kernel fork!
Since (IIRC) Merge came from SCO, and so much of SCO is now Caldera, maybe
Caldera would be the most logical partner in all this. Then again, TurboLinux
also has close ties with SCO. And RedHat is the 800 pound gorilla. Get any or
all of those three to work with NeTraverse, and many good things could come of
it.
Bottom line, NeTraverse will likely not be able to keep releasing their own
pre-patched kernels forever. The present patches are not nearly portable
enough. It is clear none of us like this situation, including NeTraverse.
Something has got to change, or Win4Lin is a dead duck (penguin?).
Ponder this: Could it be time for a "Win4BSD" product? I'm thinking of going
with OpenBSD on my main SMP workstation for several reasons, and I'd sure like
to have Win4Lin there as well. The Linux kernel politics are starting to get to
me, and I have work to do. From what little I know so far, it looks like it may
be easier to integrate Merge with BSD than it has been with Linux.
-BobC
_______________________________________________
Win4Lin-users mailing list
[EMAIL PROTECTED]
https://lists.netraverse.com/mailman/listinfo/win4lin-users
