Al Hopper wrote:
> On Thu, 27 Dec 2007, Alan DuBoff wrote:
>
>
>> On Wed, 26 Dec 2007, Roy T. Fielding wrote:
>>
>>
>
>
>>> Allowing all of the core to vote is the only way
>>> that outside contributors will become part of the process.
>>>
>
> But, in most, as in 98% (Wild Ass Guess), of the cases, a community
> putback can't be blocked if it is technically sound and meets the
> tough reliability/quality (etc) standards imposed by the ARC. Only a
> technically flawed proposal can/should be blocked (sometimes
> temporarily, until it is "up to spec"). And the entire technical
> community, regardless of company affiliation, should help to resolve
> any technical deficiencies identified by technical review to ensure a
> successful and timely integration.
>
There is an important distinction here. ARC reviews *architecture*. A
completely garbage implementation may be architecturally sound, and pass
review.
A number of the other rules, such as code quality, coding style
guidelines, test requirements, and code review guidelines originate
elsewhere. Within Sun, the ultimate enforcer for these rules is not
ARCs, but CTeams. (In fact, the ARCs derive their power from the
CTeams, which insist that projects must go thru the ARC process. A
CTeam can elect to overrule or ignore the ARC altogether. And there are
consolidations within Sun which historically get little or no ARC review.)
The OpenSolaris *project* doesn't have a well-defined place for CTeams.
Currently, the "Core Contributors" perform this role, by controlling
access to code repositories, but the whole CTeam process seems to have
been largely elided in OpenSolaris. To date it hasn't mattered, because
in practice we have been leveraging/honoring the Sun CTeams. But when
the code truly goes outside SWAN, we'd better have something in place.
The other problem, is that the various community groups largely don't
have consolidations of their own. So we have a fundamental problem here.
For ON, the problem is easily solved by making the ON community's CCs
the folks we would normally have on the CTeam, and by that membership
instating the existing CTeam rules. I don't think that has happened
yet, but maybe it has and I just missed it.
There are a few other gaps yet to be filled as well. For example, OSR
(Open Source Review) is currently performed to make sure that all
contributions from external sources are suitably licensed. Sun *has* to
have this in place, but I think the "project" also needs to have a
something here as well... a checklist at RTI time, which can be used to
call out anything non-standard (such as the Atheros HAL, which is binary
redistributable only for various reasons) needing further review by
legal "experts" to make sure that the product itself doesn't inherit any
surprise encumbrances.
Anyway, just a few things to think about... especially for the people
chomping at the bit to remove Sun from the loop here... there are
services it performs currently where there is no suitable community
replacement ready at hand. I hope we won't just ignore this issue.
-- Garrett
_______________________________________________
driver-discuss mailing list
driver-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/driver-discuss
