John Plocher wrote:
>
> The major actions that occur in a development environment are
>
> o source code is compiled into binaries
> o binaries are grouped and transformed into packages
> o packages are collected into distros and repositories
> o packages are installed and upgraded by users
>
> These need to be done within a community/ecosystem that has the
> following postulated values:
>
> o We wish to maintain a single source tree,
> o We wish to have a single OpenSolaris ecosystem, and
> o We wish to end up with a single style of packages and
> repositories that can be installed on any OpenSolaris
> compatible Distro
postulate: [v. pos-chuh-leyt; n. pos-chuh-lit, -leyt]
"... a fundamental principle."
I think that most of what you wrote is self-evidently true.
However, as far as "We wish to end up with a single style of packages and
repositories that can be installed on any OpenSolaris compatible Distro" ...
I think that if such a principle is to be upheld as a a "fundamentaly
principle" for the community, then "the community" should be involved in
defining what such a "single style of packages and repositories" looks like.
That is not happening at the moment.
What I see happening currently, is that a group of sun employees are making
their own vision of what a packaging system should look like, and then using
their considerable internal sun influence to make it "the new 'community'
packaging standard", by means of their control over what ships in Indiana.
Completely bypassing ARC, or any *community* based *process* to determine
what the requirements are, and what the implementation looks like.
Yes, they take "feedback" to some degree, but they have completely bypassed
any kind of proper ARC review of what they are doing.
And the overall Sun attitude seems to be, "that's just fine; Indiana is a
sun product, so the sun people driving it, can do whatever they like.
They're sun employees, so we 'trust them'".
This is about as anti-community as you can get.
Example: If IPS was led, and written, by non-sun employees, this would not
be happening.
Now, on the one hand, I actually agree, "Indiana is a sun product, so when
it comes down to it, it is up to sun as to what goes in there".
However, when it comes to OpenSolaris AS A WHOLE, it seems inappropriate to
me, to have the packaging methods for any and all opensolaris distributions,
to be dictatated by a set of unilaterally Sun-made choices.
Choices that ironically, arent even being guided by Sun's own ARC type
process review.
It doesnt seem like the ARC is ever going to require any kind of ARC review
of IPS for Indiana in any way(or that they have any power to do so). It
seems like a "done deal", that Indiana is going to be using IPS, in whatever
shape and form the IPS team feel like they think is most appropriate.
Is this the way that OpenSolaris process is going be from here on in?
Is it going to be, "whatever ships in Indiana, is going to be 'the standard'
for OpenSolaris" ?
If that's the case, I guess we, the solaris community, can stop bothering to
pay attention to the public ARC stuff, and just wait and see whatever ships
in Indiana, as "what we need to follow".
I'm hoping, that this is not the case, and that ARC is going to take a more
active role in the packaging and delivery side of opensolaris now.
> If you invent your own packaging system, you are implicitly also
> buying into the task of creating, populating and maintaining a
> parallel repository universe that will only be usable by a subset
> of the community.
That is certainly true.
However, this is the way that REAL "community" voice and growth happens.
If someone develops their own packaging system, it will of course initially
only be used by "a subset of the community".
The longer-term view, is to wait and see whether that "subset", eventually
becomes "the majority of the community", by the community itself choosing
whether or not to use it.
If it is a better solution, that will happen. If it is not, then it will
wither of its own accord.
The question is, is Sun going to encourage that "natural selection" process,
or is Sun going to mandate "do it our way, we know what's best for you"?
Even if Sun doesnt explicitly say that... if ARC fails to define a
binaries->packaging general API for opensolaris, then you are doing that
implicitly.
