> I think that some of the changes introduced since 2.0 were too big for a
> 2.0 release.
Hear, hear. I was shocked to find out that the version number was
actually being set at 2.0. I thought, at the very least, we should stay
at 1.4.
The 1.3 series had not really, _REALLY_, been anything close to stable
since about 1.3.54 or so, maybe before that. I remember when procps
broke around 1.3.56 or so (the first of many such breakages), and I was
relieved when 1.3.58 was announced as the "code freeze" release. We had
just gotten past that libc-make thing, and then the procps problem, and
it was high time to actually sit down and clean up the code.
Why was there never any "code freeze"? If I'm not mistaken, the ISDN
support was added just a week or so later. Since then there have been a
ton of things added or changed in the kernel, and most of them haven't
been adequately tested. A short list includes - the addition of a number
of new SCSI and IDE drivers, the addition of some new IP options, the
addition of - get this - bridging code. Was the bridging code so
important that it couldn't wait a couple of months? Near the end of
1.3, Linux had more features and options than most new cars.
We've even been adding code into the 2.0 releases (the merging in of the
FreeBSD NCR53c810 driver, which hasn't worked for me and a number of
friends, for starters) - a release that, according to the various FAQ's &
README's, should be a stable, production-quality kernel.
This is all pretty pointless, I guess. It's not like we're going to go
back on the numbers and start over at 1.4. But I think we should learn a
few things from the past development cycle. The 1.0/1.1 cycle was
actually very well-organized, if I remember correctly. All the bugs were
out by 1.0.9, and the 1.1 series froze about 1.1.70 (I think), and there
was an actual freeze and plenty of bug busting. 1.2 was a release that
almost everyone was proud of, and everyone agreed that it was ready for
release.
Let's try to make the 2.0/2.1 series more like the 1.0/1.1 series, and
less like the 1.2/1.3. We're already at 2.0.10; that alone makes me
worry about the true stability of this release.
shag
Judd Bourgeois | When we are planning for posterity,
shagboy@bluesky.net | we ought to remember that virtue is
Finger for PGP key | not hereditary. Thomas Paine