avalon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Berin Loritsch <blorit...@apache.org>
Subject Re: [Proposal] HiveMind Service Framework
Date Thu, 13 Nov 2003 12:32:32 GMT
J Aaron Farr wrote:
> Quoting Leo Simons <leosimons@apache.org>:
>>But I'm weary of two things:
>>1) community seperation. ...
>>2) subprojects don't work well for avalon. ...
> I agree with Berin and Leo that the larger issues are community related rather
> than technical.  I find that unfortunate since I would like to see Avalon as a
> center for all things COP/IoC/SoC related, taking in and developing 
> many different ideas.  My biggest concern with Hivemind in Jakarta vs Avalon is
> that both Avalon and Hivemind would miss out on a lot of synergy that would help
> both projects, but if Avalon's not ready for more autonomous subprojects (which
> has been the case in the past), then, as Leo pointed out, it may be best to have
> Hivemind in Jakarta with the option of joining Avalon open.  If not now, then I
> hope at some point Avalon will reach the maturity of being able to take on and
> foster projects such as Hivemind.

I think this also begs the bigger question of what's the difference between an
"autonomous" sub project, and a TLP?  The fact that we have a PMC that manages
a (currently) unknown piece of code does bring up certain management issues.
What happens when, on the off chance, that another IP related issue comes up for
HiveMind?  There is only so much the PMC can do when they are operating in the
dark.  A PMC operates most efficiently when they have a reasonable idea of the
projects being managed.

A second community related question comes in with tooling these containers.
Currently we have three containers, with one or two destined to be obsoleted.
Yes, there is (and should be) a lot HiveMind and Avalon can learn from each
other.  In fact, I would very much encourage it.  In order for an effective
exchange of ideas, we have to map the concepts that apply in both contexts.
In HiveMind we call XYZ function "foo", and in Merlin we call it "bar".

There is a lot of theory discussed, which influences where we go.  There will
be a ramp up time so that we are all talking the same language.  While I would
like to spend time to see what similarities there are, what the pros and cons
are of implementation details, but I just don't have the time.

I think the proper solution is have someone hire me full time to work on this
stuff ;P.

All joking aside, for the two communities to mesh well, folks from both
backgrounds can't be overprotective.  There would have to be a lot of consensus
building, and some ideas pushed off for a while.  It's hard enough building a
community around one component spec, it would be even harder to build a
community around two seemingly very different specs.

To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org

View raw message