Keith M Wesolowski wrote:
> I fail to see how deferring a solution until adequate information is...
> "self-destruction". Could you elaborate?
Many people in and out of Sun are trying to find a way to make these
things work on OpenSolaris ASAP. Blastwave, SunFreeware, Sun Engineering,
Nexenta and the other distros are all pushing hard towards this goal.
To date, the ARC's reaction to this stream of proposals has been to
choke on them. 320+ emails for the star fasttrack, 100+ for unison,
etc.
If the ARC starts rejecting all these projects without clear and
actionable guidance about what /exactly/ these projects can do now
to make their stuff available, the need for architectural effort and
understanding will be ignored or bypassed and the bit will be defacto
set that OpenSolaris projects do not need to be ARC reviewed.
In other words, we will have the moral high ground, but we will end
up being irrelevant and ignored.
Whether this is done by an irate Sun executive or by a distro that
decides to go its own way or a repository site that throws open
its doors to anyone with anything to offer, no strings attached,
the result is the same: The idea-development-delivery chain will
no longer effectively have any requirement for ARC interaction
for any project effort.
By creating an optimized fast path thru the ARC review process that
works for these projects, we end up with a process that retains ARC
interaction as an integrated step:
idea-ARC-development-delivery
The ARC review for this class of project is simply delegated to the
project team. This is subtly (but critically) different than "nobody,
including the project team, is expected to understand and manage the
architecture of their project".
> I don't share your commitment to this goal, but I will express a
> requirement
It sounds like your fears are pretty close to mine: it doesn't
matter if you like or hate the "unreviewed, underreviewed, or
exception-stuffed content", the bug is that we are stuffing it into
our source tree without any mechanism to differentiate it from
the "overreviewed and tightly constrained" traditional Solaris
Enterprise stuff.
I added a reference to the your comments and the related Companion CD
consolidation to the wiki page. I agree that, if all these components
were aimed *there*, we wouldn't be having this problem.
-John
