> On Thu, May 31, 2007 at 09:22:20PM -0400, Ian Murdock wrote:
>
>
>> So, it seems the crux of the matter is the following decision:
>>
>> 1. OpenSolaris should remain a source base only. Sun
>> and others use that source base to build (potentially incompatible)
>> operating systems based on the OpenSolaris code base.
>>
>> 2. OpenSolaris should be an operating system in its own right.
>> Multiple implementations (distros) can still exist, but they must
>> remain compatible with each other to use the name OpenSolaris.
>>
I'm not a member of the OGB, but I play one on TV.... :-) Well, not
really, but I did *run* for OGB... so I'm going to voice my thoughts
anyway.
It appears to me that item #2 can be broken down further. The idea of
having a reference distribution is totally different from the
requirement to be compatible.
There is some benefit from compatibility guarantees, as the binary
compatibility guarantee in *Solaris* (not the Open one) has shown.
There may also be some benefit from having a reference distribution.
(Right now SXCE serves in that capacity.)
But those two things, are, I think orthogonal.
I'd love to have a compatibility guarantee done enforced by some kind of
test/conformance suite. Conceivably it might even be possible for an OS
with an entirely different kernel to earn the right to be branded, if it
could pass the conformance tests. (Ala the UNIX/POSIX conformance
tests.) Of course, I don't think that anyone here has the resources to
burn on creating, agreeing upon, and testing for conformance.
I think having a reference distribution is also useful. But, as others
have pointed out, start by developing the distribution, and then get
consensus that it should be called the reference distribution. But
start from a working base (which might openly be developed with that
end-goal), before seeking some kind of blessing for the brand. At least
that is, I think, a better approach.
-- Garrett
