Re: [picocontainer-dev] Trying an experiment to kill off ContainerRecorder (Nano)

Instance components for session level is not something I thought of  
with the ThreadLocal/Storing thing.

If application level, then regular caching is the strategy

Request level, in-theory, a no-caching and no-storing CA would still  
cache instance registrations.

Where do you use instance registrations dude ?

- Paul

On May 21, 2008, at 7:41 AM, Michael Rimov wrote:

>>> Here's another patch.  This one builds and passes
>>> tests, but I think
>>> the tests need work - things are quite different
>>> now.
>>>
>>> We need to ensure that only one container is made
>>> per scope ... for
>>> the entire life of the web application.
>>>
>>> I'd love someone else to take a look. Someone who's
>>> familiar with the
>>> NanoWar architecture.
>>
>>
>> This someone must be me...  I will throw a look
>> to night. Container recorder is only needed to speed
>> up
>> initializing ofleaf containers - so if there  a better
>> solution it can go for good.
>
> Hi Konstanin,
>
> Hey, perhaps we can do a combination of storage + the new  
> ReusablePicoContainer
> for the leaf containers.  (I've never envisioned it for anything but
> Request-level Containers anyway).
>
> Paul has a "HttpSessionStoring" behavior solution, but we need to  
> flesh it out
> for handling instance registrations, but if we get around that  
> hurdle, I'd be
> happy too. :)
>
> 							-Mike
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe from this list, please visit:
>
>    http://xircles.codehaus.org/manage_email
>
>


---------------------------------------------------------------------
To unsubscribe from this list, please visit:

    http://xircles.codehaus.org/manage_email