On Jun 16, 2006, at 7:08 PM, James Carlson wrote:
I'm a bit confused about what you want. Achieving those standards
means a high degree of attention to detail. It's not just grabbing
the latest and greatest source, typing 'make', and tossing it over a
wall. There's a _large_ amount of work that goes into making those
amazing things, and having an open source base to draw from doesn't
somehow cancel out the other issues. (As has been said before, free
software isn't free of cost.)
I'm asking about a compromise. Keeping software reasonably up to
date, at the same time not so bleeding edge as to it coring all the
time. I understand it is an undertaking, but you have a lot of
resources to draw from in this very community. I.E. Blastwave, SFW.
They have this process pretty much down, some slight modification,
some tweaking here and there, and some additional testing cycles, and
you'd be all set. Stable software, reasonably up to date, and a
*part* of Solaris. That's what people look for, myself included. If
you haven't seen the thousands of posts around the net of disgruntled
people who tried out Solaris based on the technical merit of ZFS/
Dtrace/etc, and were amazed at how easily they all worked, and how
well - then when they decided to get php5+modules up and running - oh
- have fun. ;)
That work takes time, people, and effort. It means figuring out how
the software may need to change to fit with the rest of Solaris. It
means coming up with test cases. Exactly the things it sounds like
you don't want us to do. After all, if we were to do that, then you
would get something like /usr/sfw -- software that's out of date or
otherwise unwanted because it has gone through an intentional and
painstaking design and review process.
There is a balance to be found..
So, which do you want?
I want (and lots of other people want) reasonably up to date software
that is stable and monitored for security issues. Basically, what
Blastwave offers (or SFW). Is our (OSOL) community position to simply
make a kernel, and a very basic userland, and let people slap
whatever they want on there, however they want, with no best-practice?
Personally, I *like* having blastwave and sunfreeware being
independent and robust. They give me exactly what I would have if I
were to run Linux -- easy access to the newest things, all pre-
compiled
and ready to run, but no deeper guarantee that they're going to play
well together or not just dump core unceremoniously. In many cases,
that's "good enough," and it certainly mirrors the experience you get
elsewhere.
Yeah, ok, you're not seeing eye-to-eye with me I think. I'm saying
OSOL should have something like blastwave (if not blastwave itself)
integrated into solaris, with a few more layers of testing slapped on
top. That'd be the best of both worlds, easy access to relatively new
things, all pre-compiled and ready to run, and a deeper guarantee
that they will play nicely together and not dump core. Why settle for
good enough? Is that the community stance? "It works, who cares if we
could do something better - I don't need it so let's just stick with
it." We'd all be riding horses across the open plain if that were the
case. There is *always* room for improvement, and something as core
to the success of the OS I would think should be a priority, because
right now it's a major detractor. Go look up any of the OSNews posts
about Solaris/OSOL this or that. The primary complaint is always
software packaging. People want to throw on the OS, and be ready to
go. There is no reason why we should not have something like that. I
don't understand the logic behind retarding OSOL's development simply
because we have "good enough."
I think the remaining issue is a contractual one: you seem to want to
pay Sun for support of open source software that's actually developed
elsewhere. In that case, contact your support folks. I don't know
how they draw up agreements or evaluate costs (I'm not involved in
that), but perhaps it's a business model that could be made to work.
No, that wasn't my position nor intent to get across. My apologies
for my miscommunication. My point is, the software management/etc
should be integrated into OSOL (and thus eventually Solaris) and have
a "best practice" sort of methodology to it. Just like *every* other
server OS short of Windows. Well, Slackware is kinda iffy I suppose.
There is a reason they are all doing this, and there is a reason why
Solaris (and conversely) OSOL take a back seat to these other OSs.
Let's make this clear now, it is NOT technical related issues. Nobody
even dares say anything negative about ZFS/Dtrace/Zones/etc anymore,
it's generally accepted they are superior to just about everything
out there. Think about it - if our OS (OSOL) is the best technically,
what other reasoning might there be that it is lagging behind other
OSS OSs? Usability. HW support used to be an issue, usability is all
that really causes a problem now as HW support is quickly getting
flushed out. I can't think of a better place to start improving
usability then to make relatively current software easy to install/
manage/update, for a *normal* user. You've got the people to do it -
they *already are*. You just need a bit of QA polish on the top, and
you'll be set. What is the big hang-up about integrating this? It
seems like there is some mystical black cloud floating over Blastwave/
SFW that everybody is afraid of at Sun, yet they constantly preach
for us to use it - they just won't integrate it. Ask just about any
admin how they get software up and running on Solaris - most will say
Blastwave/SFW. Very very few compile everything by hand anymore. Why
do we keep it segmented off? This is OSOL, it is a community project.
So let's involve the community, and maybe Sun can get involved too.
You think?
In any event, that doesn't sound like it's really related to this
mailing list. This list is for discussing Open Solaris, not the
support or marketing of any particular distribution, including Sun's
Solaris. I don't think the issue of whether Sun offers support for
third-party applications (of any kind) is an Open Solaris topic.
I know what the list is for, and OSOL is more than a kernel to me.
Did I miss something here? We had a lengthy discussion about this
earlier on this list, concerning the relationship between Sun's
Solaris, OSOL, and Solaris Express. OSOL is what becomes SX, is what
becomes Solaris. This has been beaten into my head. The eventual goal
from my understanding is to have all of Solaris opened, and a part of
OSOL, and from that point on, Solaris will be built from OSOL, with
Sun's QA/etc "polish" on top, or however you'd like to call it. So, I
think very much this discussion is part of OSOL. I use Solaris/OSOL
interchangeably, because at some point they are pretty much the same,
one just lags behind to pick up additional testing/stability, and
contains some non-OSS code. I'm sorry I ventured in my original mail
to discuss Sun support, I just feel that Sun is a huge part of the
OSOL community, and seeing as OSOL is the basis for what Solaris
becomes, Sun might want to have a supported software system. Just
like RH, Suse, Ubuntu, etc. Again, as I mentioned before, a balance
needs to be found. Obviously, it isn't Sun's/OSOL community member
code everytime when you're dealing with OSS software. That doesn't
mean you can't support it. If it's good enough that _everybody_ (Sun
employees included) tells people to use it, it should be good enough
to integrate with OSOL. I don't know why there is such heavy
resistance to doing so. I can't speak for Dennis, but if I were in
his position, I'd jump at the opportunity and embrace it fully. It'd
help OSOL be accessible to *normal* users, and it'd go a long way to
improving usability. If you spend just a little time surfing the net
and reading comments from first-time users trying out Solaris, you'd
realize this is a huge stumbling block to mass acceptance. I hope you
can see that - but this is a community project, and my intentions are
for further community involvement and the bettering of OSOL itself. I
hope Dennis will chime in with his wisdom here, because he knows a
heck of a lot more than me on what's involved in something of this
scale, and he's bound to know what users think about the situation,
as between him and SFW, they *are* the packaging community, already.
Certainly not Sun, I suspect you can count the number of people using
the stuff in /usr/sfw for more than five minutes until they get
Blastwave setup on one hand. This should signify something that needs
improvement.
I don't want OSOL to become another Linux, as in a "we are a kernel
only" type community project. I love that people are making distros
to suit their needs, but it sure would be nice if the default
"distro" of OSOL was all inclusive (and thus Solaris...) and usable
for most tasks out of the box. It's silly (to me at least) that most
*experienced* Solaris users/admins are having to go out of their way
to get Blastwave setup within the first 10 minutes of getting the OS
installed, after fixing the silly niggles that need to be worked out.
It's just an extra step for no reason as far as I can tell, and if
this is truly a community project, I don't know why we are keeping a
large part of the community away from being a *part* of OSOL. Make no
mistake, Blastwave really is a big chunk of what makes OSOL/Solaris
usable for people.
Again, my intent is to discuss things pertinent to OSOL, please do
not mistake this for some contractual-whatever discussion. I poorly
voiced my opinions before, I hope this has cleared it up for you. We
may never see eye to eye on this aspect, but it is a community
project, and as a part of the community I felt it was something
necessary to discuss. Oddly enough, it's all been discussed before
(obviously not specifics - but in general) , and the general
consensus has always been something needs to be done about usability
for people who haven't touch it before (new users) and that software
is a major point of contention. My apologies if I went about this the
wrong way.
Respectfully,
David
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
