Ian Murdock wrote:
The divide here seems to be that OpenSolaris started life as "Solaris'
kernel.org" as opposed to the "community version of Solaris". So,
rather than creating a single binary distribution to begin with,
Sun released the source code to the kernel and some other pieces above
the kernel, built a "distribution" around it called Solaris, and
encouraged (or at least didn't discourage) others to build distros too.
I don't think it was a mistake, per say, but it simply represented the
strategy of the time based on the circumstances of the time (although
Solaris wasn't built around OpenSolaris; OpenSolaris has always been a
subset of the code in Solaris).
We are talking about almost three years ago now. So many things were
different at that time. Solaris 10 hadn't even shipped during the time
we were planning OpenSolaris. Also, we've had to release the Solaris
code in pieces over time, and my understanding is that we simply
couldn't release it all at once and we couldn't offer a binary distro
based exclusively on the open code (though we would have liked to) due
the pieces we couldn't (and still can't) release (though the engineers
can correct me on that).
And yes, we wanted to encourage the formation of distros, and some who
are saying that was a mistake now never said a word back then. In fact,
the notion of, or the hope of multiple distros, was shared by many
across the company and in marketing and engineering and in the
community. We wanted people to take the code we had released and
experiment with it. To do things we hadn't done. This is similar to the
naming of OpenSolaris. People say it was a big mistake to tie
OpenSolaris to Solaris like we did. Well, we did that intentionally,
actually, since we felt we needed to deep brand recognition of Solaris
to help OpenSolaris, and we didn't like the experience of some of the
other open source communities with separately branded projects. And
really I can't remember any of the current critics voicing concerns back
then about the name.
Instead of saying what went before was a mistake, why can't we build on
what has gone before? Or will the decisions we are making now be
characterized as a mistake next year by someone else? I'd love for Indy
to be successful in whatever form it takes because I think we can reach
out to engage quite a large number of people in emerging markets. Some
of the work is already going on in other projects, so we have a good
opportunity to grow the project here.
Jim
--
Jim Grisanzio http://blogs.sun.com/jimgris
In my view, that was a mistake. First of all, I'm not sure what
having multiple distros buys us. Some here have suggested that
multiple distros is what made Linux succeed where others (e.g., FreeBSD)
didn't. I disagree--Linux took off before there was any concept of a
distro (I know, I built one of the first ones, and it was done
to accelerate something that already had traction). The reason
Linux took off had more to do with the BSD lawsuit than anything
else. There were other reasons too, but that was the main one.
I _do_ know the downside of having multiple distributions:
incompatibility. The vast majority of the world sees Linux as a single
platform, but in reality it's a collection of about five _mostly_
compatible platforms. Mostly compatible doesn't cut it in the world of
SLAs though. The problem isn't users wondering "Where do I download
Linux?" as some have correctly pointed out is a non-issue. That was the
problem we solved in the early '90s when we created the first distros.
The problem today is, say, the ISV wanting to target "Linux". Ok.
What distributions? Keep in mind that from a support point of view each
distribution costs as much to support as a wholly different OS. _Mostly_
compatible doesn't cut it. I'm very surprised this argument
doesn't resonate better around here. Do we really want that for Solaris?
-ian
_______________________________________________
indiana-discuss mailing list
[email protected]
http://opensolaris.org/mailman/listinfo/indiana-discuss
