[OpenAFS] Re: [OpenAFS-devel] 1.6 major feature: pthreads = on

Russ Allbery rra@stanford.edu
Thu, 17 Jun 2010 15:57:30 -0700

omalleys@msu.edu writes:

> Pthreads appears to be the only major feature in the 1.6 release.

I'm not sure where you got that impression, but this is not the case.
There are several different features in the 1.5 series, but pthreads is
not one of them; OpenAFS already supports pthreaded file servers in 1.4.

> It needs to be on by default since you are -calling- it a major release
> and people -expect- some issues. You might as well have them with your
> MAJOR feature.  Otherwise it is just a lame fork of the 1.5.x branch you
> might as well just call it 1.5.x instead of renaming it 1.6. It isn't
> like the lwp code is yanked out. If people want 1.6 with lwp they can
> recompile.

I think several people are confused about the version numbering scheme
that's used in AFS, so let me try to clear that up.

OpenAFS uses the version numbering scheme that the Linux kernel was using,
namely that odd numbers are development releases and even numbers are
stable releases.  1.5.x releases, with the exception of Windows, are not
intended to be stable releases and are not deployed by the average user.
1.4 is currently still our stable release.

Therefore, to determine whether it's worth releasing 1.6, you should look
at the changes relative to *1.4*, not the changes relative to 1.5
versions.  In essence, the entire 1.5 release series are alpha, beta, or
release candidate releases for 1.6.0.

Just since OpenAFS 1.5.38 -- in other words, not even the entire release
series of 1.5 -- there have been 522 changes worth mentioning in release
announcements, the vast majority of which have not been pulled up to 1.4.

Even if DAFS were not in 1.6 at all, there's absolutely no question that
we've accumulated enough changes to warrant a 1.6 release.  Given that
DAFS will also be included, I don't agree with your argument at all.

Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>