Keith M Wesolowski wrote:,
> I assert that integration adds value but mere aggregation, generally,
> does not.
I actually agree with you :-)
If there was a place to put "aggregated things" that didn't want to
pay the price to be integrated (i.e., doing the Solaris way -vs- the
Linux way), then the level of ARC interaction/involvement goes way
down. It might even go away completely if there was a way to make
it a "self review" sort of decision process.
OTOH, if things wanted to move/be moved from the "experimental" or
"aggregate" repositories into the "tightly integrated into Solaris"
one, then there is plenty of room (and need) for ARC engagement.
Setting expectations: I expect there to be 100,000 things in
the "experimental" repository, 10,000 in "aggregate" and 100 or
so in "integrated". As Jim says, the cost of transforming s
given something from "aggregate" to "integrated" is pretty high,
so we need to carefully pick and choose. (The difference between
the "experimental" and "aggregate" repos is probably just a
function of who can decide to push packages into the repo:
uncontrolled -vs- controlled).
> The best opportunity for compromise was offered by the Companion
> consolidation, which could have been used as a vehicle for unfettered
> aggregation of arbitrary (and arbitrarily toxic and badly-integrated)
> software, while the other consolidations continued using strict
> quality-focused processes.
I don't think the idea has been lost, just misplaced :-)
I'm pretty sure that there is a truism in here somewhere that says
that stuff bound for the "experimental" or "aggregate" repositories
should not go into ON (or SFW?); the problem is that today, we don't have
any alternatives. Maybe the OS.o Project mercurial repos can be made
into "Consolidationettes" that feed directly into the "experimental"
IPS repository...
-John
