avalon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stephen McConnell <mcconn...@apache.org>
Subject Re: @avalon
Date Sat, 22 Mar 2003 21:13:50 GMT

Peter:

Thanks for you reply - see notes in line:

Peter Royal wrote:
 > On Saturday, March 22, 2003, at 12:33  PM, Stephen McConnell wrote:
 >
 >> I would argue that the process would be more constructive if we
 >> made the assumption that a certain set of tags exist - and from
 >> that assumption, work on ensuing that the respective
 >> interpretation of those tags by different tools and resulting
 >> meta-info are consistent.
 >
 >
 > The primary stickler seems to be over the lifestyle tag being in the
 > @avalon namespace. I see Pete's argument, only things that *ALL*
 > containers support go into the @avalon namespace.

In practice all of the containers - Fortress, Merlin and Phoenix make 
assumptions about lifestyle.  In the case of Merlin and Fortress there 
are mechanisms to support different assumptions based on information 
supplied by a compoent type (whether declared in a marker interface of 
via formal meta-info descriptors).

If component assumes a "transient" lifestyle - it will not function 
properly within Phoenix.  To overcome this, Phoenix needs an ability to 
recognize that a particular component type requires a particular 
lifestyle strategy (independently of ability to support it). The 
recognition of @avalon.lifestyle by Phoenix is a first step by dealing 
with recognition.

This isn't about "stuff that all containers support".
It's about stuff that all containers can recognize.  This is important 
point.

 > As a compromise, ow about we establish an @avalon-x namespace that
 > holds items that, in the future, may be promoted to the @avalon
 > namespace?

The downside of this approach is that it leave Phoneix in isolation. 
Components documented to require a particular lifestyle will only be 
functional (with confidence) within Fortress and Merlin

 > That way Fortress and Merlin can both use @avalon-x (or @x-avalon) for
 > the lifestyle tag, and phoenix would just ignore it. If/when the point
 > in time comes that all containers are able to handle such components,
 > we could promote the tag to the @avalon namespace.

If we apply this approach from here on - a common Avalon container 
architecture is in effect impossible.  That process involves active 
collaboration and compromise.  For example, while the set of tags that 
Berin has proposed are workable relative to Merlin, they are not 
necessarily what I would choose in all cases - however, that's a small 
price to pay relative to the benefiot of moving forward.

 > I see you & Berin's point Stephen, and I see Pete's. Both are valid.

If we were talking about this as "Phoenix must support these lifestyle 
semantics" - then Pete's aguments would have some validity.  But we are 
not asking that.  What we are asking is that a very small number of tags 
be addressed - each representing a semantic notion that exists in *all* 
containers.  All this does is establish recognition that the semantic 
exists.  Pete's argument is that the semantic does not exist at the 
level of a type. I think Pete's argument is wrong - and this can be 
easily shown in the Phoenix code (Phoenix applies the same semantics to 
all component types) and equivalent type level handling in both Fortress 
and Merlin.

The real issue here is what should happen when a container or tool 
encounters a type level criteria it cannot support.

Cheers, Steve.



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


Mime
View raw message