Hmmm... Maybe I should start back peddling on being down on James now that
all of this comes to light.

I might need to collaborate with HiveMind when it comes time to try my hand
at providing an extension layer. (I hope the word extension makes it a
magical effort , but more likely than not it will end up looking like a lot
of paper clips and masking tape holding together a very fragile structure :)
)

So, no hard feelings right James? This sounds do-able? ;)

On 7/31/06, Howard Lewis Ship <[EMAIL PROTECTED]> wrote:

I'm keeping the package names seperate, to facilitate splitting out
Tapestry IoC into a standalone project, ala HiveMind. Right now that
is not a priority.

The way I've been layout out the Tapestry IoC APIs, it should be
possible to create a kind of bridge RegistryBuilder that can read
HiveModule XML and make those appear (largely) like native Tapestry
IoC modules and services.  There's a bunch of things that don't line
up however:

Decorators (service interceptor factories in HiveMind) can now match
(i.e., target) multiple services. This makes them more like AspectJ
pointcuts.

Configurations in HiveMind are standalone. In Tapestry IoC each
service *may* have *one* configuration, as an unordered collection, an
ordered list, or a map. This turns out to be more than sufficient (you
can always create a service that exists to vend out its configuration,
and thus simulate multiple configurations per service).

Object ordering is more robust in Tapestry IoC:

http://tapestry.apache.org/tapestry5/ioc/order.html

Tapestry IoC doesn't yet have a number of HiveMind features:
- Registry shutdown
- Pooled service lifecycle model (just "perthread")
- Object providers (
http://jakarta.apache.org/hivemind/hivemind/ObjectProviders.html )
- Symbols

Many features of HiveMind won't be needed in Tapestry IoC as they can
be more easily accomplished in Java code.  Thus "instance:" or
"class:" object provider prefixes are useless.

On 7/30/06, Benjamin Tomasini <[EMAIL PROTECTED]> wrote:
> I have been using HiveMind now for quite a while and have come to depend
> on it quite heavily on many projects - some web apps, some not.  It's
> features have very much improved the quality and modularity of my code.
>
> What would prevent a user from employing Tapestry IoC as a general
> purpose container like HiveMind?  Would there be any benefit to building
> one monolithic Tapestry jar and a set of segmented jars (something
> Spring does) for more targeted use?  Or why not keep Tapestry a single
> project with sub-modules for IoC and the Web framework?  Alternatively,
> what about merging Tapestry IoC and HiveMind at some later date?  It
> could remain as HiveMind, or it could take on the Tapestry IoC name.
> The conversations on this list indicate that key principals of both
> projects intend to keep some degree of parity anyway.
>
> Ben
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>


--
Howard M. Lewis Ship
TWD Consulting, Inc.
Independent J2EE / Open-Source Java Consultant
Creator and PMC Chair, Apache Tapestry
Creator, Apache HiveMind

Professional Tapestry training, mentoring, support
and project work.  http://howardlewisship.com

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




--
Jesse Kuhnert
Tacos/Tapestry, team member/developer

Open source based consulting work centered around
dojo/tapestry/tacos/hivemind.

Reply via email to