On Fri, Jun 08, 2007 at 02:50:36PM -0700, johansen-osdev at sun.com wrote:
> > It might be wasteful, since some concerns dispensed with earlier are
> > likely to resurface.  But still, sounds like a good compromise to me.
> 
> I disagree with the statement that previously mentioned concerns have
> been dispensed; rather, they've be avoided.

I don't think so, though I too would rather dispense with closed review
altogether.  The key thing here is that commitments would have to be
done in public.

Now, there is the issue that it is possible to commit on inception --
clearly that should not be possible at closed inceptions.

But if this closed-inceptions-ok proposal is aimed at making it possible
to ARC early and ARC often without giving up confidentiality until close
to the end of the project, and there is a sufficient strong real or
perceived need for such confidentiality, then I think that's a fair
compromise.

> | I would expect an OpenSolaris C-team to disregard any invisible, closed
> | review, whether made within Sun or elsewhere, when determining whether a
> | project is ready for integration into an open consolidation.  That is,
> | if your project is Sun-confidential, it is Sun-confidential all the way
> | through and past integration, and you are therefore targetting a
> | Solaris-specific consolidation that is irrelevant to this discussion.
> | Or, if your project is targetting an OpenSolaris consolidation, you are
> | obliged to engage in open review from a sufficiently early point that
> | everyone has the opportunity to participate and to meaningfully advise
> | your project team.
> - WESOLOWSKI <20070605221545.GE472012 at sun.com>

This is addressed by the Rich Lowe/Garret D'Amore proposal: no
OpenSolaris C-team would be presented with a project reviewed only
behind closed doors.

> | [W]hen one contemplates the Nexenta example, it becomes much clearer
> | that these secret codebases have to be treated as opaque objects
> | irrelevant to and separate from OpenSolaris.
> - WESOLOWSKI <20070606193217.GA710356 at sun.com>
> 
> Nobody seems to have substantively disagreed with these statements, yet
> we continue to argue that there is some imagined need for closed review.

Well, we've not had much, if any input from anyone about the need for
closed review.  Jim's argument for why simply saying "no closed reviews"
won't do is that it will happen that project teams will try to force
integration of unreviewed code.  My view is that as soon as one team
tries it and loses (in the sense of being made to wait for an open
review and deal with any TCRs that may then issue) in a sufficiently
spectacular (in the sense that it makes for good theater) way then it
will stop happening.

> Closed projects don't belong in an OpenSolaris consolidation.  Can we
> agree and move on?

No.  Clearly until a project can operate privately at least until comes
to the ARC for the first time.

The issue is when to force de-cloaking.

That could be: (a) before any ARC case(s) is(are) filed, (b) at ARC case
filing time, (c) at ARC inception review time, (d) at ARC commitment
review time, (e) at c-team inception review time, (f) at c-team
commitment review time, (g) at integration time.

(a) is best.  Anything later than (c) is unnaceptable (at least it
appears to be for all who've said anything in this thread).  You think
even (c) is unacceptable.  I'd prefer (b) but find (c) livable.

Further, I assert that we can't force (a) since there is no discrete
time prior to filing an ARC case at which the CAB could reasonably
declare that the project had to become public (there may be such a
discrete time for many, but not all, Sun-internal projects, namely, when
the project is funded or otherwise shows up on some management
dashboard).

So we're really down to selecting either (b) or (c).

Nico
-- 

Reply via email to