David Brown wrote:
On Wed, Dec 05, 2007 at 05:07:48PM -0800, SJS wrote:
No, and I won't. You floated a big number as an example of "not
wheel-spinning", and alone, that isn't sufficient. Or even a good
indicator, much less a refutation of wheel-spinning.
Why is it any better than giving a different meaningless number out of a
bug tracking system. Without looking at the bugs you can't tell how
significant they are, so it doesn't tell you anything either.
Actually, it does. Since there is no incentive to game the bug system,
the data on closure rates, importance, activity, etc. is a useful measure.
But, to be quite honest, calling the work of current linux kernel
developers wheel-spinning is rather insulting.
Well, what would you call the whole VM fiasco? NFSv3 "locking"?
Arguing against ZFS because it "telescopes" multiple layers
(translation--our OS functions aren't well encapsulated so we'll shoot
the messenger)?
Has the Linux kernel measurably improved since 2.4? 2.6.1? 2.6.9?
*Can* you measure how much it has improved?
From my point of view, it regressed quite a bit from the 2.6.4
timeframe because something started mangling the FS drivers.
Unfortunately, only XFS had a good enough compliance suite to squawk.
After about 10 versions, FS stuff seems to have been patched enough to
make things roughly work on x86. However, the other ports of XFS were
still pretty broken and nobody really knows why. Other FS systems may
be equally broken; they just don't have compliance suites so we don't know.
I joined this list rather recently, and am trying to get an idea of what
kind of group it is. So far, all I've really learned is that it doesn't
really have kernel developers on it, and the postings almost seem hostile
to kernel developers. I'm not sure I really would want to go to a regular
KPLUG meeting, or even continue on the mailing list.
Well, this is a Linux list. That includes users as well as developers.
Users always outnumber developers by at least an order of magnitude.
Hostile to kernel developers? Not really. I tend to follow what the
Solaris, FreeBSD, OpenBSD, and even OS X developers say without too much
hostility (even with Theo de Raadt being an absolute flaming asshole).
Hostile to *Linux* kernel developers? I might fall into that category.
This has been increasingly so over the last couple of years.
For some reason, Linux kernel development is developing an infallibility
complex. "We're right, you're wrong. Our way or the highway."
Unfortunately, they're *not* always right. 2.6.X is littered with "not
right". And it doesn't look like it's converging much. There is a
reason a *lot* of businesses are not moving up from 2.4.X kernels.
Combine an increasingly complex piece of software with an infallibility
complex and you are bound to produce hostility.
However, in the interest of fairness, I'm biased. I'm one of those
silly FreeBSD users. I started using FreeBSD several years ago because
of a building sense of unease about Linux. The intervening years
haven't made me regret that decision.
-a
--
KPLUG-List@kernel-panic.org
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
