On 21/07/2010 11:01, Joerg Schilling wrote:
Robert Milkowski<mi...@task.gda.pl> wrote:
Of course IPS is included in the OSOL distribution for a long time now.
One might argue that the OSOL is a community distribution as apart from
Sun many community members contributed to it in one way or another.
Sure, most of the work has been done by Sun guys. But it is not that
much different to Fedora as I think that most of the work is being done
there by people being paid by RedHat...
I am not sure about where you live but in the real world, there is NO IPS
in Solaris. IPS was introduced in Indiana (why not in Solaris) and Indiana
is a nice proof of concept on how to slow down things by working against the
community.
I wrote about the Open Solaris distribution and not the Solaris 10.
And there's been IPS in the Open Solaris distribution for quite some
time now.
Many people have also deployed OSOL 2009.06 or its /dev branches on many
systems.
When it comes to Indiana which was/is just a code name for the Open
Solaris distribution I don't think they've been working against
community. Just because they don't necessarily agree with you doesn't
mean they work against community.
Many times there were people here who did not agree with you - does it
mean you worked against community?
The real point is that something had to be done about packaging system
in Solaris. Except for lots of talk and ideas no-one, including you, did
anything useful here.
So far Sun committed resources and actually started doing something and
while some people like you keep criticizing IPS imho most of us actually
like it. Now you are coming late to the party and except for
non-constructive whining you don't produce much more. IMHO either help
IPS or get something implemented. But I would suggest that unless you
can come-up and *deliver* something *substantially* better then IPS,
don't even bother as you would only waste your time.
> From the important political aspect of how to place OpenSolaris into a OSS
environment, nothing did happen since summer 2005:
Many, many people here wouldn't agree with you as a lot has happen since
2005 in Open Solaris space.
Except for some drivers and a small amount of code 99% of the Open
Solaris distribution is open source.
Then we have /contrib repository with many packages provided by
community (including me) - in many respects it is already better than
Blastwave and definitely more open and friendly to contributions.
- Sun/Orcale still releases a closed source sed and they even still
releasae a closed source pax and the worst: they spend effort in
adding extensions into the closed source pax that are not
based on recent POSIX standards althouth star is able to replace
the closed source pax since 2004 and includes an easy to extend
framework to implement archive extensions based on the recent
POSIX standards.
gsed is also included so if you prefer open source version then you have it.
closed-source sed is there and I can see why it probably would be a
waste of resources to make it open when gsed is available. PAX - similar
story.
star - it is a really good tool but most users don't really care. From
Sun/Oracle perspective it is a lot of effort and risk to get it
integrated for not that much of a benefit. But then if someone from the
community wants to get it integrated you can - but it is *you* who must
put some effort. It's been discussed here many times so I don't see much
point going thru it again. So far I'm still 100% convinced that if you
would put some real effort it would get integrated long time ago.
- There is still no OSS i8n code. There was a Solaris emancipation
project in 2007 but the visible results are this single line:
libc_i18n: Contact:// johnsonnenschein at gmail dot com
Do you want to give a hand? Or do you expect someone else to do it for you?
You are talking a lot about community but most of the time here it seems
what you mean by that is that you would like to be able to dictate to
Sun/Oracle what and how should get implemented and let them to do it...
But it doesn't work like this. If it is really such a big problem for
the community (which I don't think is the case) then the community
should get it fixed...
- There is no Xorg package from Blastwave, is there such a beast
from other sources besides Indiana?
I don't know and I really don't care. It is in the Open Solaris
distribution, it is completely open source and I don't see much point
having it in Blastwave as well.
- There is no set of GNOME packages from Blastwave that include gdm
and correctly support internationalization.
Does this exist from other sources besides Indiana?
Why would I care?
From my assessment, the most important task for a community distro is
to get rid of the Closed source stuff.
Once this is done (and I am planning to work on this task based on build 130),
Excellent!
While b130 is a good starting point (why b130?) please make sure you
test it with the latest source tree so it can get integrated into the
main repository. Also please, please, please make sure you do not use
all of your Schilly build platform but rather do it the osol way -
otherwise it will be mostly an wasted effort.
Why use the latest source tree if this causes lots of problems?
b130 has a clean SVr4 repository and this works together with Blastwave.
There is no community distro that is based on IPS, so currently IPS does
not really exists. There is no migration path.... you either get the SVr4
system _or_ IPS, this is not how migration works.
Well, on my notebook I currently have Open Solaris b134 distribution and
it definitely has IPS.
BTW: What is the "osol way"? I am using a build system that is superior to what
Sun introduced and I am happy to collaborate with interested people.
It doesn't really matter if it is superior or not. If you want to
collaborate with some project you need to be able to play by its rules.
You can't expect all different communities changing their build
infrastructures to something else just because someone thinks he has a
better solution. There are also many other factors which need to be
taken into account.
But frankly - instead of sending emails, adopt your build platform so it
works with all or at least most open solaris consolidations, show that
it is better and that it is worth migrationg, make sure all the other
tools around distribution work with it properly and then you have a
chance to actually make it happen. Or are you again expecting someone
else to do it for you?
When it comes to the "osol" way - check http://opensolaris.org - there
are many documents there on how to contribute to it. People have done it
(including me) successfully so it is possible and it is not that hard.
An interesting aspect is also that people at Sun did signal that they are not
interested in collaboration. Instead of collaborating with people that already
created a distro, they hired new people that did not have skills on a Solaris
distro. Now that most of the related people are no longer at Oracle, does this
give us a new chance? There currently is silence, so we currently need to asume
that there is still no interest in a collaboration with the community.
Well, maybe you reading these signals wrong way then? Others read it
differently and have been able to collaborate with Sun.
When it comes to people leaving Sun/Oracle - unless you have some
internals knowledge and can actually prove it it is just your
speculation. And if I were to speculate on it I would have to disagree
with you.
If you are really thinking about a community driven distribution then I
would argue it would be much better to take as much as we can from the
OSOL distribution and enhance on it instead of starting from scratch for
no real reason.
Good point, so why do you believe tha the Indiana distro ignored this?
Ignored what exactly?
closed-bins is a separate issue and has not much to do with creating a
distribution per se - fixing the issue by re-writing closed components
and getting them integrated into the main tree would fix the issue for
all os distribution.
This has been discussed many times, once Snoracle demonstrates that there
is a real interest in a collaboration, it makes sense to start thinking this
way. Currently there is no such signal and it is therefore important not
to waste time to implement things nobody is interested in while there is a
method that does the same with less effort.
Can you point to an instance when you or someone else tried to get
integrated an open sourced replacement for some part of closed-bins and
it has been rejected by Sun with no real basis?
I doubt you can.
When it comes to "signals" - actually if you search archives you will
find some Sun guys trying to encourage others to get it done. What other
kind of a signal do you need?
The real question is what is missing (no source available) in order to
build OSOL distribution the way Sun does?
Is this important when we do not get any signal from Snoracle that
there is an interest on a collaboration?
It definitely wasn't the case.
Now it is slightly different and we don't really know where exactly we
are but hopefully it will all clarify soon.
--
Robert Milkowski
http://milek.blogspot.com
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
