ARRRRGH! Guys, who's breaking -STABLE's GMIRROR code?!

Colin Percival cperciva at freebsd.org
Sun Sep 10 13:25:58 PDT 2006


Michael Abbott wrote:
> Damn, I'm confused now.  Let me try and get this straight:
> 
> CURRENT
>     This is, by definition, broken a good part of the time, and is what
> it says, namely current, ie work in progress.

Yes.

> STABLE
>     This is broken some of the time and .. uh .. isn't really all that
> stable, actually.

"STABLE" means "you can update FreeBSD along this branch without needing
to recompile applications or kernel modules".  This means that companies
can ship binary drivers for their hardware and say "this driver will work
on FreeBSD 6.x" (which isn't possible on Linux).

The fact that there are occasionally bugs introduced... well, that's an
inevitable consequence of the stable branches being development branches.

> RELENG_n_m
>     This is completely stable and only tracks security fixes.

Security fixes and "critical errata".  The requirements for something
being committed to such a branch after the release are that:
1. It must be an important bugfix, and
2. I must be absolutely certain that nothing bad will ever happen as a
result of someone updating a FreeBSD n.m system to the latest updates on RELENG_n_m.

> RELENG_n (RELENG_6 at the moment)
>     Has somebody just said that RELENG_6 = STABLE?

Yes.

> I'm going to guess then that RELENG_7 is CURRENT.
>     No, this doesn't make sense to me at all.

RELENG_7 doesn't exist yet.  RELENG_7 will be 7-STABLE once it exists,
some time in 2007.

Colin Percival



More information about the freebsd-stable mailing list