On Jul 21, 2004, at 2:34 PM, Joshua Tinnin wrote:
Then why do I hear that 5.2.1-RELEASE is not ready to be called STABLE?

FreeBSD's -CURRENT tree has generally been reasonably stable, but there have been periods (including quite recently with threading/#define PREEMPTION) where -CURRENT has not been reliable enough to qualify as -STABLE.

Why would it be downgraded? Why have there been no STABLE 5.x branches? Or am I
just confused?

There have been no 5.x branches which qualify as -STABLE, correct.

You may be confused, but it is the result of the extent of changes to 5.x taking longer to settle down than the developers would want. The hope was that 5.1 or 5.2 would be stable enough to promote 5.x to -STABLE perhaps six months ago. This hasn't happened, and is the reason why there is a big push to get 5.3 stabilized and solid.

Again, there is some leeway for a .0 release, such as 5.0, to not be as stable as the earlier 4.x releases, but the extended period where 5.1 and 5.2 were put out as RELEASES while 4.x remains -STABLE has not been helpful to users trying to determine what the best release for them to run should be.

Almost exactly one year ago, I wrote:


Matthew N. Dodd wrote:
[ ... ]
I'm still not sure we should be encouraging new features in -STABLE;
additional hardware support and bugfixes are one thing...

Doesn't the term "MFC" refer to a change or new feature that has already been added to -CURRENT, and is under consideration for being backported to -STABLE because the change is important, of general interest and utility, etc?

If the question is "when should new features not be merged back into 4.x", my response would be that should happen after 5.x is tagged as -STABLE and 5.x is being actively recommended for to all users including newbies, not just early adopters. If the concern is "is it better to spend time trying to get 5.x -STABLE then it is to spend time on 4.x", well, that makes perfect sense to me.


PS: What does not make much sense is 'releasing' a 'new version' of software which is not intended for the end userbase to actually use.

Attempting to reduce the scope of problems with a .0 release is a noble goal, but good intentions can be taken too far. If a user asks "what version should I run" and the answer isn't "the latest release", well, that indicates a problem. If a release candidate isn't expected to be better than the prior numerical version for the end users, then the release candidate isn't ready.

Perhaps I'm drifting off-topic a bit, but I remember administering Sun machines during the transition from SunOS 4.1.x to what marketting called Solaris 2.x. Sun didn't do itself or anyone else a favor with SunOS 5.0 through about 5.5; it wasn't until Solaris 2.5.1/SunOS 5.5.1 that Sun's customers got something significantly better than a .0 release, or (perhaps arguably) better than the prior major version. That really sucked, people, so please excuse my vehemence.

[ Or don't. If the comparison between SunOS 5.x and FreeBSD 5.x earns me flames, rabid criticism, and the undying emnity of whomever, so be it. :-) ]

[EMAIL PROTECTED] mailing list
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Reply via email to