[oi-dev] OpenIndiana and illumos

[Garrett D'Amore]

On Thu, 2010-11-18 at 14:15 +0800, Christopher Chan wrote:
> On Thursday, November 18, 2010 02:09 PM, Garrett D'Amore wrote:
> > On Thu, 2010-11-18 at 03:32 +0100, Jesus Cea wrote:
> >> OI has provided hope for some of us. OI team already explicitly said
> >> that the plan is to move to Illumos. Having "something" filling the gap
> >> between OpenSolaris and a future Illumos based release is keeping people
> >> interested in the technology instead of flying away in flocks.
> >
> > I understand that, although there have been releases since, and b147 is
> > available today.  Plus there is SX11.  Continuing to burn cycles on b148
> > that could have been better spent addressing the real future needs is
> > what bugs me.  And more to the point, I'm being constrained from making
> > improvements I need to (in illumos-gate) by lagging OI.  And most
> > grievously, it is OI's pseudo-compatibility goals with Oracle's releases
> > (and now with its own prior releases) that create some of the biggest
> > handicaps.
> 
> What constraints are you having?

There are several dependencies which are tangled.

For example, I'd like to clean up the removal of some closed code, like
Solstice Enterprise Agent.  I can't do that without modifying net-snmp
in the SFW consolidation.  Which means I either create my own fork or
start working with OI.  But the changes would be incompatible with
Oracle Solaris (and ONNV 148.)

Other recent examples include perl and some packaging related stuff.

I'd like to integrate libm into ON.  (I'm not sure whether this is a
problem with OI or not, but it seems like something that ought to be
*coordinated* at least.)

I'm also looking at stuff relating to OpenSSL dependencies.

I'd also *really* like to help fix the g11n problem but I can't do that
in OI today.  The reason is that the legacy Oracle depends on private
interfaces in Oracle's ON consolidation, which illumos doesn't have.
(Illumos has the same capabilities, but the *interfaces* are different.)

Ultimately I have been spending a lot of time trying to avoid making
life any more difficult than necessary for folks running OI and onu'ing
to illumos-gate.  But this continues to get increasingly difficult for
*me*.

> 
> 
> >
> >>
> >> So, OI has a role.
> >>
> >> What do you want from OI people?. If they are only half a dozen
> >> "integrators", what kind of help you want from these poor guys?.
> >
> > I want to help them.  But I can't do that while they are stuck trying to
> > follow in Oracle's footsteps.  *If* we can move forward, I can help them
> > and probably *double* their resource.  If we can't do that, then I'll
> > probably wind up spending that resource starting a new project instead.
> >
> 
> Where do you want to take an Illumos based distro?

I have several things I'd like to do:

a) I want an illumos micro-distro.  This is needed most especially since
we don't have a reasonable alternative in the form of OI or anything
else today.

b) I would like a desktop distro.  OI could be it, but I think an SXCE
work-alike with SVR4 packaging and a clean upgrade from S10 would be
incredibly compelling.  If I have to give up on OI, then this is the
direction I'll go.

c) Of course, I also need to continue to support our own Nexenta
distributions, which use Debian tech.

	- Garrett