On 9/21/05, Howard Lewis Ship <[EMAIL PROTECTED]> wrote:
> I want to pursue the idea of doing more of the service construction in
> non-reflective code.
>
> So, a module will be able to delegate to a class, and that class will
> be used instead of a ServiceImplementationFactory to build services;
> there will be a naming convention used to define what services are
> defined in the module class (vs. in the XML).
>

(I just reread your original proposal and the discussion about this:
http://thread.gmane.org/gmane.comp.jakarta.hivemind.devel/1388) As I
understand this idea, this would allow me to define a module using a
Java class instead of an XML descriptor. This would indeed be quite
nice! Although I believe we would need an API which both the XML
parser and this new "Java parser" used to register the module (and
it's contents) with the registry. I spent some time to outline a
proposal for this API on the Wiki:
http://wiki.apache.org/jakarta-hivemind/DescriptorApiRevampProposal.
Did someone get time to look at this?

> I'm hoping this will help speed application startup.  Tapestry 4.0 is
> too slow to start.
>
> I'd like to look elsewhere for other speedups; perhaps a custom Map
> implementation tuned for small numbers of elements.
>

Sounds interesting. Have you run a HiveMind application through a profiler?

> Annoations would be nice, but I think we want to keep
> HiveMind/Tapestry useable with JDK 1.3 for a while longer ...
> something to discuss.  Perhaps JDK 1.4 so we could ditch ORO.
>

The annotions could be optional, no? IMHO moving to JDK 1.4 would be a
good thing.

> I'd like to see a replacement for BuilderFactory called
> "InjectionFactory" that uses significantly streamlined XML.
>

+1

> I want to be able to use a Schema to parse an arbitrary XML stream.
>

Would this result in configuration contributions or what's the usecase?

> I want to see descriptors that allow Spring code to be used as native
> HiveMInd services.
>

That would be nice. Do you think it could involve some Java coding to
make the services more HiveMind like? Where would we host these
descriptors?

> I'd like to (expirement with) caching the contents of XML descriptors
> on the file system so that restarts of an application are faster.
>

Also sounds interesting.

--knut

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

Reply via email to