The Old Way Was Better
Gary W. Swearingen
underway at comcast.net
Mon Sep 8 16:09:24 PDT 2003
Doug Barton <DougB at FreeBSD.org> writes:
> The decision to delay any kind of release in the 5.x branch was the
> right one to make. The fact that it's still missing features that we
> wanted to have 2 years ago should give you an idea of why. :)
I agree, except that with good naming there should be no problem with
releasing buggy and incomplete code very early. (I'm talking in
general there, and not sniping at 5.0.) The problem has to do with
what code is referred to as "X.0". Because of human nature, it either
needs to be the buggy incomplete alpha/beta code or code that is
announced as the real thing, ready for wide use and recommended for
use (by most people) over anything in RELENG_X until "X.0.1" or "X.1"
is released. And also because of that human nature, if you really
want to maximize the number of people using the first good version,
you'd better not call it "X.0" because most people expect it to be
buggy. The way 5.x was released was just as people tend to expect
software to be released. As I said before, I'm not prepared to
critize what happened with that. My criticism is of the
confusing/misleading use of CURRENT, RELEASE, and STABLE.
> As for the rest of your post, it's all very interesting, but incredibly
> unlikely to happen. The creation of the RELENG_4_X branches solved the
> immediate need for a "stable branch plus security fixes." 5.x is still
> -current, and while we do need to be more careful with our marketing
> (and more careful with what goes into a 5.x release), massive branch
> renaming just isn't going to happen, nor is expanding the number of
> branches going to help.
I agree with all of that, except the implication that I suggested
any expansion of the number of branches.
More information about the freebsd-chat