Alan Coopersmith wrote:
> So that's a bit more discussion than I expected. For those who want to
> allow closed cases to integrate into OpenSolaris code, please offer a
> proposed policy that explains how this could be handled, and why it
> wouldn't
> just become carte blanche for anyone to integrate without ARC review?
I think Richlowe had the idea, that closed _inception_ should be
permissible.
A final open commitment should be required.
I've also suggested that the "closed" reviews should not meant that they
were held "internal" to any contributor, but rather that only the
OpenSolaris ARC members and the case owners were part of the review, and
that the case materials are locked up (only accessible to ARC members,
case owners, and maybe interns.) Note that these ARC members could be
employees of Sun, Joyent, or Microsoft for all I care. That shouldn't
matter.
Again, instead of thinking of "closed" as Sun-internal, think of it as
"ARC approved" individuals only. Note that this may require NDA or
something to be dealt with prior to the case going to ARC. It may also
require ARC members to sign a blanket NDA agreeing that ARC-confidential
matters will be held in confidence. Already other open source
communities have this for "privileged" members of their communities. (I
had to sign such an agreement when joining NetBSD, for example.)
Further, the final (open) commitment review can be held without
requiring all the discussion that occurred during inception to be open,
I think.
>
> What gives Sun special rights to integrate without public review that
> wouldn't
> also allow [apologies in advance to those about to be named, who are
> committed
> to not going down this path] Roland to declare that ksh93 has some
> secret he's
> not allowed to reveal to us in ARC review and thus he held closed
> review at the
> Roland's Ksh93 and Komodo Dragon Supplies, Inc. ARC, or allow
> OpenSound to put
> in whatever they want, including breaking all existing Solaris Audio
> apps,
> because they have secret driver interfaces and held a closed review in
> their
> private ARC? If we allow any project to claim secrets allow them to
> have ARC
> review by a closed body they selected, how do we have any meaningful
> system-wide
> architecture?
>
See above. Stop thinking of closed review as something that happens
inside a contributor, and instead something that happens within the ARC
community, with doors closed to the greater public.
-- Garrett
