On Jul 11, 2005, at 3:52 PM, Geir Magnusson Jr. wrote:
On Jul 11, 2005, at 2:56 PM, Dain Sundstrom wrote:
The console is neither. Since these are very different code
bases, I think they need to be addressed differently:
Console:
We bring the code directly into the geronimo/trunk/sandbox. We
work on the code there, and any people that worked on the code
before the donation, contribute via patches. Once the code is
ready, we move the code to /geronimo/trunk/applications.
ORB:
We bring the code and programmers into the Apache Incubator as a
subproject supported by and destined for Geronimo. We develop the
initial code an community in incubator, and then bring it into the
Geronimo project with a separate SVN location. Once the project
develops a good community of it's own we move the project to a top
level project (this could take several years).
These two solutions are not in conflict.
I hope I didn't implied they were. I am simply trying to say that
they are totally separate situations.
The problem is that IIRC, the consensus for the ORB wasn't to do it
in incubator, but bring the TriFork code and people here and close
and involved directly in what we are doing.
Ah... what? Just because the code lives in incubator doesn't mean
that we can't work closely with the trifork guys. I think it is
clear that everyone wants to work with the trifork team closely.
I have no problem with what you say above, but we should treat all
contributions the same way, and a contribution from the Incubator
is the same as from outside, is it not? Whatever process we
require of individuals to get commit status is the same?
I'm actually happy if your answers are "no" and "no" as long as we
clearly define our process.
I guess you're not going to be happy. I think that we have different
situations here. My guess is every donation will be a unique
situation. We need to measure the situation and act accordingly.
Note: I perceive both of these code bases as special cases and
not precedents. The console is specific to Geronimo and really
doesn't work without it, so it belongs in Geronimo.
Well, these are precedents to see how we bring code in (as more
will be coming and yes, some of it will be very specific to
Geronimo). Hypothetically, if TriFork offered their EJB container,
then it - how OpenEJB works notwithstanding - is not a standalone
project because the EJB spec can't be implemented legally outside
of the full container, is therefore Geronimo specific, and belongs
in Geronimo.
I hope no one would do that. That would be incredible damaging to
our community. How would you feel if Trifork donated their web-
service implementation? We could suck it into Geronimo and get
everyone using it. Of course that would really hurt Axis.
I think we avoid any situation that would undermine an existing
healthy open source community. If someone wants to donate something
to compete against an existing healthy Apache licensed open source
community, we can simply suggest they work with the existing
community or start a new one.
The ORB supports a large specification without a (healthy)
existing Apache licensed open source version. If there were an
existing apache licensed open source ORB, I would rather see the
code donated and worked into an exiting project. Alternatively,
the group donating the code could start a new project outside
Apache, and develop a healthy community of it's own. I do not
think that Geronimo should ever assist in undermining an existing
(healthy) open source project.
That's fine, but I don't think the donators wish to go this way at
first, and I think that we're happy to accommodate them.
What? That was a hypothetical situation. I wrote "If there were an
existing apache licensed open source ORB", but as I see it there is
not one, so we should a new project and community here.
-dain
