Noel J. Bergman wrote:

An article on The Server Side
(http://www.theserverside.com/home/thread.jsp?thread_id=22371) started me
thinking again about HiveMind today.  The major issues mentioned there are
the same as mentioned by Howard below:



I've looked to see how we could graft HiveMind into Avalon and vice-versa, but they are really quite different beasts. The type-1 vs. type-2/type-3 split is intrinsic and difficult to reconcile. HiveMind's concept of a module doesn't map so easily into the Avalon space, and HiveMind's free-for-all approach doesn't jive with Avalon's dogmatic security model
(including its explicit application construction descriptor).



Are these really incompatible concepts?


My impression is that Howard's assessment of Avalon is based on the Phoenix container. Howard is perhaps unaware of that over the last year or so a lot of development has been going on that is resulting in the establishment of a set of common containment facilities. These include automated assembly, composite component management, component meta-models, deployment frameworks, development and management tools, etc.

I think Howard's point concerning the type-X issue is non-trivial - but not because of any insurmountable technical issue - instead, its much more a question of maintaining clarity and consistency within a component model. I fully expect to see the Avalon platform incorporate these approaches - but in a way that is complete and consistent.


Isn't Merlin, for example, trying to do more with auto-assembly?


DIR="LTR">


Assembly management is a facilities included in the avalon-activation package. It provides support for automation of assembly processes, eliminating the need to explicitly declare consumption/production relationships. The focus of attention should really be on the facilities - Merlin is just an example of a container that leverages these features extensively.


Doesn't our model allow us to push services to be used rather than having to pull them from a service manager? I really don't see why this is an either-or issue.

Its not an either-or issue. Both push and pull models of service acquisition are possible, preemtive or dynamic, etc. This is an area that is being worked on at the moment. I would be interested in hearing from some the HiveMind guys about their thoughts on this area as its a subject I happen to be addressing with respect to the Turbine components.


Likewise, free-for-all vs security can be viewed as a policy.

I suggest that people give some thought to how TO collaborate with HiveMind
and Avalon.  I think that the combination of people and technologies could
be great.

It would certainly be interesting to get together with the HiveMind guys perhaps over on [EMAIL PROTECTED] and push some ideas around. As a minimum it would provide a chance for the Avalon crew to learn more about HiveMind content and ambitions and for the HiveMind crew to catch up where Avalon is today and where it is going.

Cheers, Stephen.

--

Stephen J. McConnell
mailto:[EMAIL PROTECTED]




--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to