Re: freebsd 7 release process

On Fri, 25 Jan 2008, Aristedes Maniatis wrote:

> How can non-committers like myself keep up to date with the progress of 
> freebsd releases? I read stable, current and cvs mailing lists. I've even 
> looked at the qa mailing list but it looks completely abandoned (perhaps the 
> links to it should be removed from the web site).
>
> I see this page:
>
> http://www.freebsd.org/releases/7.0R/todo.html
>
> but it tells me that we are just waiting for Robert Watson to complete 
> testing of the MAC framework changes. Nothing in commit logs or mailing 
> lists correlates to that so I assume it is incorrect.
>
> Really, all I want to do is to get a handle on the known outstanding bugs, 
> but the FreeBSD bug tracking system is very crappy (compared to almost all 
> other open source projects I use or am involved in) and has no way to look 
> at bugs by reported-in version or target release. So my question is: how do 
> other people know what is happening? Does re have a mailing list somewhere I 
> can read?
>
> I know many people are putting in vast amount of effort here, so this is not 
> a criticism of the delays. Just without any regular bulletins, emails to the 
> mailing lists, bug tracker or web site updates I can't tell what bugs are 
> left, what are fixed and what I need to be aware of.

Thanks for the prod -- I've update the todo list to reflect the 7.0RC2 issues 
I'm aware of (that is, the things currently blocking building and shipping 
RC2).  If I bump into any others, I'll add them.  It takes a bit for the web 
site to rebuild, so I've attached the current pending TODO items for 7.0RC2 
here:

                                              A race in TCP will lead to very
                                              occasional returning of
                                              0.0.0.0:0 rather than the
    TCP accept                 Gleb Smirnoff, correct IP address via the
    race condition In progress Robert Watson  socket address returned by
                                              accept(). A patch is in HEAD and
                                              RELENG_7, and will shortly be
                                              merged to RELENG_7_0 for
                                              inclusion in 7.0RC2.

                                              Races in the sleepq
                                              implementation will occasionally
                                              result in sleeps with timeouts
                                              failing to wake up. A patch is
    sleepq race                               in HEAD and will shortly be
    condition      In progress John Baldwin   merged to RELENG_7 and
                                              RELENG_7_0 for inclusion in
                                              7.0RC2. The sleepq code is being
                                              reviewed for other possibly
                                              related bugs and further fixes
                                              may be forthcoming.

                                              Multiple race conditions in the
                                              tty/pty code have lead to
                                              occasional panics. One set of
    pty race       In progress Konstantin     problems has been fixed, but
    condition                  Belousov       others are still being explored
                                              It is aniticipated that the
                                              remaining problems will be fixed
                                              for 7.0RC2.

There have been status e-mails from Ken ... but only to the private developers 
mailing list.  I've asked him if he could start also sending them to 
current_at_/stable_at_, it was just an oversight that they weren't being sent there.

We're extremely close to a 7.0RC2, which with any luck will be the last RC, 
btw.  7.0 is looking like one of the most stable ".0" releases we've ever 
done, and has some very nice features...

Robert N M Watson
Computer Laboratory
University of Cambridge