[OpenIndiana-discuss] LX Branded zones

[Jacob Ritorto]

On Fri, Jan 23, 2015 at 5:30 PM, James Carlson <carlsonj at workingcode.com>
wrote:

> Having lived through the effort at Sun, I don't think it's a great idea.
>  The problem with the LX branded zones is that you need a huge team of
> people to chase every little feature in Linux.  The bizarro world /proc
> that they have ("hey! let's use /proc/kitchen/sink!") is by itself a
> giant resource suck, and that's just one "small" component.
>

  Thing is, in the real world, Linux libs can't (or won't tend to) move
that much w/o the fear of breaking libc compatibility.  Lx emulation is a
pragmatic way of getting all (well, not all, but many) Linux junk to run on
unix.  If, perchance, a lot changes in their universe and people start
writing app code to require it, then we do the same in tandem, but that's
most decidedly not what's actually happening in the wild right now, so it's
not really that much to babysit.  Just accept that it's never gonna be
perfect, because the nature of their whole system philosophy / nature of
the beast is never gonna be perfect.  So stop being a 'correctness'
cathedral scientist for a sec, stfu and fudge it like they do.  We (pulling
from the smartos peeps, perhaps, because they have to for business reasons)
maintain it by doing what we have to to keep the apps happy -- and just run
that shit with no porting.

  Gross, disgusting bazaar mentality, yes.  But quite timely and useful.
Opens a whole new world of running slipshod, albeit hip and popular, code
on a nice os.  And the kicker is that you can instrument it better than
they can on their own platform, which, in an admittedly snarky way, is kind
of amusing and has a whole lot of legitimate adoption opportunities and
implications.

  Man on the street says we need to get this into Illumos rtfn.  Remove it
later if it turns out to be utter shite.

-j