Re: URGENT: A Plea to the owners of

Todd Graham Lewis (
Wed, 2 Apr 1997 18:50:02 -0500

On Wed, 2 Apr 1997, Theodore Y. Ts'o wrote:

> The problem is that none of us has time to respond to every single
> lame-brained idea that gets raised. Ah, for the good old days when most
> of the "I've got a great idea" was usually followed (in the same
> message) with a "and here's a patch which implements it".
> We still have some of that, but there are a lot of people who seem to be
> posting on the list simply aren't kernel developers, and don't have the
> expertise to do kernel development. Sigh.....

Indeed, a whole lot of people (Todd says, assuming in true bourgeois
fashion that everyone else is just like him) have been drawn to
programming through their involvement with Linux. I for one have made
one- or two-line modifications to a kernel and _actually gotten it to
work_! Thus armed with confidence that even mere mortals can do
kernel development, I like to watch the pros at work, in the hopes of
someday being able to post one of these "I've got a great idea and here's
a patch" messages.

At this rate, though, I'll never figure out how to do kernel programming.
Back in the Good Old Days(tm), I could read along and discover "Wow, you
mean that you can't use floating-point operations from withing the kernel
because the registers get clobbered? Boy, it sure is great that I read
that in l.d.k. instead of finding out on my own like the NTFS guys did!"

Speaking with Miguel de Icaza at Usenix, I listened as he bemoaned the
slow pace at which developers are coming up to speed and participating in
kernel development. We should treat this as a real problem. When Pedro
was hired away by Cisco, who picked up the networking development? David
and Eric Schenk, both of whom already do a lot of work. Eric is a prof,
and has limited time; David is an undergrad. What are we going to do when
David graduates and Microsoft offers him $10^6/year to make NT as fast as
Linux? Who will take his place?

If Linux kernel development has to go underground, and it already has to a
large extent (viz. Alan's recent comment on one of the private lists that
its purpose was to protect the developers from the (people) on comp.os.
linux.networking), then new developers will become unable to absorb the
Tao of Kernel Development, and this is a bad thing.

I for one have resolved not to propose any more of my ideas for kernel
development until I can back them up with patches. I decided this the
last time I proposed a (patch-less) idea and Ted pulled out his standard
"Where's the patch?" spiel. Not having any formal training, programming
or otherwise, it's taking me a while, but I'll get there eventually. (I
was a political theory major, for goodness sake; cut me some slack.) Not
being able to read the development discussion which used to saturate this
list is a real hinderance to my effort in this respect.

Again, I'm not saying that we should discourage users with questions. Far
from it, I too have done my time in the trenches, answering 40 or 50
questions on the help groups/lists in a single evening. I'm not even
saying that moderation is the answer; Dave Barr had another idea, and I'm
sure others do, too. I'm not even saying that I'm particularly clueful,
David. 8^) But I am saying that we should make a real effort to keep
this list concentrated on kernel development.

I'm sorry if people feel that meta-kernel-development discussions are
inappropriate, or that I put this topic forth in the wrong way, but what
I've seen recently is very disheartening, and I think it does little good
and much ill to the linux effort to allow the present trend to continue.

I'll crawl back into my lurker shell now; sorry for having disrupted the
important development efforts going on.

Todd Graham Lewis             Linux!                 Core Engineering
Mindspring Enterprises   (800) 719 4664, x2804