by Since 1.1.93 was released, numerous people have posted copmlaints about how
their systems have suddenly become much too swap-happy.
Others have responded by forwarding a message that Linus sent to the
linux-kernel mailing list about this problem, usually followed by something to
the effect of, "If you're going to run the latest development kernel, you
should be subscribed to the linux-kernel list. Don't expect the latest
development kernel to be stable. If you're not willing to subscribe to the
linux-kernel list, then you shouldn't post problems with the kernel."
I would like to remind those people of something. Somewhere in the 1.1.80's,
Linux declared a code freeze. He said that from that point on, only bug fixes
would be introduced into the kernel. He said that the thought the kernel was
quite stable. And he encouraged everyone to run it, because (paraphrased),
"If you don't run it and therefore don't discover a bug in it that affects
you, that bug will make it into the 1.2 release."
I do not subscribe to the linux-kernel mailing list. Although I've submitted
a few Linux kernel bug reports and patches in my time, I really don't have the
time to actively keep up with it, given all the other volunteer activities on
the Net in which I participate (and given my full-time job and preparation for
a year-long Sabbatical in Israel :-). However, I got the distinct impression
from Linus' code-freeze announcement that he wanted as many people as possible
to use the new 1.1.x kernels from that point, even people who do not subscribe
to the linux-kernel list. I'm sure many other people got that impression as
well.
We've all seen how well the "code freeze" was enforced. Just like the last
time Linus declared a "code freeze" and said that only bug fixes would be made
after that point, significant functionality changes have been allowed into the
kernel since the alleged freeze. Recall that the last "code freeze" mucked
things up so much that Linus decided to cancel it and not to do a 1.2 kernel
release as he'd planned.
Linus, when you declare a code freeze and tell people to try out the kernel so
that bugs can be found and fixed before the "user release" (or whatever the
even-numbered releases are called), then you owe it to the people who are
willing to do that pre-testing for you to ENFORCE THE CODE FREEZE. No new
functionality. No "tweaks" to the swapping code which allegedly improve it
while in fact making it worse for many people. No new CD drivers. No new
files for Alpha and Sparc support. No new configuration parameters which
force people to rerun "make config" after virtually every patch. NOTHING
EXCEPT BUG FIXES.
For those of you who are about to respond to this message with, "Hey, Linus
doesn't get paid for his work! You've got no right to make demands on him!"
-- save it. I don't get paid for maintaining XRN, either, but the people who
use it would be mighty miffed, and rightfully so, if I repeatedly declared,
"This is the final beta release before the next production release! Only bug
fixes will be incorporated from this point on!" and then proceeded to release
ten additional releases with new functionality in most of them. No, I can't
make "demands" on Linus and the other kernel developers or declare that we
have the "right" to expect them to operate a certain way, but I can ASK that
they operate in a certain way, and I don't think what I'm asking for is
unreasonable.