Matthieu Riou wrote:
On Thu, Sep 11, 2008 at 12:02 AM, Bernd Fondermann <[EMAIL PROTECTED]>wrote:
Matthieu Riou wrote:
On Tue, Sep 9, 2008 at 9:45 AM, Stefano Mazzocchi <[EMAIL PROTECTED]
wrote:
Thorsten Scherler wrote:
I don't know what others think, but I could imaging small, dedicated
labs going more or less directly to other projects as sub-projects.
Yeah, I would like to hear the opinion of Stefano especially and the
people that created the bylaws to start with.
A lab *has* to do thru incubation, even if it's just a formality. If
not, somebody willing to bypass the incubation procedures, could simply
pay a committer to pretend it was his/her code and just work on it for a
while in labs until good enough and then just ask for TLP.
We have not cleared any IP here, labs wasn't created for that and we
don't know how to do it (and don't care to care about it at this stage).
Labs have been around for a little while now, shouldn't we start caring to
lift the no release constraint? The way I see it we have:
* incubator: takes care of community building and IP issues.
* labs: none of the above.
Incubator and Labs make not much difference, when community building is
concerned. Incubator currently has more attention, while in Labs it's easier
to add a new (already-ASF) committer. Labs has one mailing list, Incubator
feels like it has a few hundred. Both is an advantage/disadvantage, it
depends on how you look at it.
But the fundamental setup is the same. They are both ASF projects,
following its community building standards.
I beg to differ. The incubator has actually pretty high community standards
for graduation (AFAIK the lab has none as you transition to the incubator)
and the only-ASF policy of the lab make it impossible to attract someone
from the outside. That's two very different dynamics.
Contribution from the outside is possible, of course, as in every other
project. JIRA and the mailing list are your friends.
This is all just a matter of how the responsible PI handles it. With one
or two sustainable outside contributors I as a PI would hurry my lab
into incubation and make them proper committers.
I really don't see any other issue than to actually find those
contributors ;-)
What people do not make enough use of here at Labs is that they can publish
non-ASF releases, just like outside projects do. Why? Because only very few
lab codebases have reached a level of maturity where a release is making
sense. I, for example, will make a private release of Vysper as soon as both
underlying RFCs are reasonably covered.
Right and as it's an outside release, you don't benefit from any of the IP
policies that the ASF has in place. You're certainly free to follow them but
from a user standpoint it's moot.
If there is a lab which wants to do a release, and there is only one
committer working on it, I would not want this code to be a fully-vetted
ASF release (be it here in Labs or elsewhere) since this software lacks
community and user support. So the fallback is to make a non-ASF
release, or to not make a release at this point in time.
If instead there are enough committers working on a lab, they would be
best served to go into incubation. There they would not only have their
own list (reducing the load on [EMAIL PROTECTED]) but also be able to do a
proper ASF-release (yet "-incubating", I admit) and eventually become a
project on their own.
I actually don't even understand why people bother with labs for non
strictly apache internal stuff. The initial idea was that other Apache
people would notice through the labs mailing list and get involved. So far
I
haven't seen much evidence of that beyond "sounds interesting". So if it's
only to provide a SVN repository, not so interesting. I rather github,
thank
you.
So if labs don't bother about IP, what's the benefit both for the
foundation
and lab users?
It's much easier to eventually go into incubation.
Evidence?
No evidence yet, just a feeling. Let's wait and see what's becoming of
Droids. :-)
Infrastructure is ASF standard.
A lab is started by a ASF committer, so I'd expect a lab to be above
average in terms of quality than the average github project.
So do I, specifically in terms of IP. Which is why I find the no-release
policy annoying because it might be a low hanging fruits. As we're all ASF
folks, we could take care of reviewing each others releases.
Basically, yes. But just think about why the Incubator makes 3 mentors a
requirement nowadays. It's because mentors should feel more connected
and responsible to actually review a release than the rest of the IPMC
folks. Before that, some podlings had big trouble getting releases out
of the door because they could not aquire a quorum. This has improved
with mentors, but is still an issue sometimes.
Bottom line: Releases are best done at the Incubator. They know how to
do that. Labs doesn't.
Labs has the attention of interested ASF folks.
A lab has a dramatically higher probability that I will eventually take a
look at it than the most interesting github project, because I simply don't
have the time to review all the projects on every open source platform.
I personnally don't have the time either to review every TLP + subproject +
incubator project + lab that we have. The probablity is high for sure, but
it's still fairly slim :)
+1
Bernd
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
