Salut,
John ROM wrote:
> Hi Alexey,
> as always I have just to much work at work but this weekend
> I did spend an 2 hours working with
> Grizzly 2.0.
>
> Just 2 comments or questions.
>
> 1)
> Maybe keep the MemoryManager Interface really slim?
> maybe witout the util wrap methods in there. For example when integrating Ken's Slab Allocation Pool
> I am only interrested in allocate( int size ) and maybe int maxAllocationSize();
> I don't want to have to be forced to implement wrap(..)
I'm also tempted to agree with you :-) Alexey is on vacation for another
week, so let's wait for him :-)
>
> 2) Grizzly 2.0 has 3 Abstractions :
>
> org.glassfish.grizzly.Transformer,
> org.glassfish.grizzly.filterchain.CodecFilter,
> org.glassfish.grizzly.filterchain.Filter
>
> Please correct me if I am wrong,
> so there are two API's which handle read /writes.
> A Filter API which is like Grizzly ProtocolChainFilter API (selectorThread,XXXexecute Methods,WorkerThreads) and a (let me call it) TransFormer API (Connection.read/write)
>
> My feeling is that CodecFilter should not extend Filter so
> that there would be no need to implement Filter. This would also make it clearer that
> the processing is very different.
> Or have I missed something? Maybe CodecFilter needs some specific FilterChain functionallity
> (other then List) ?
>
But we will need an interface for Codex-like filter, right? What this
interface should looks like?
A+
-- Jeanfrancois
>
> Many Greetings
> John
>
>
>
>
>
>
>
>
>
>
>