Mailing-List: contact jackrabbit-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: jackrabbit-dev@incubator.apache.org
Received-SPF: pass (asf.osuosl.org: local policy)
Message-ID: <42B81F1D.7070309@zitting.name>
Date: Tue, 21 Jun 2005 17:07:25 +0300
From: Jukka Zitting <jukka@zitting.name>
User-Agent: Mozilla Thunderbird 1.0.2-1.3.3 (X11/20050513)
MIME-Version: 1.0
To: jackrabbit-dev@incubator.apache.org
Subject: Re: ObservationManager for JCR-RMI
References: <42B7CD44.3070700@day.com> <42B7EFA3.8050300@yukatan.fi>
 <42B806CA.1030502@day.com>
In-Reply-To: <42B806CA.1030502@day.com>
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: 7bit

Hi,

Felix Meschberger wrote:
> Ok, I only used the Channel class from the concurrent library. So I 
> implemented a simple replacement based on a LinkedList.

Nice, thanks.

>> - Out of curiosity, did you think about solving the listener problem 
>> by using a remote listener interface instead of the polling solution? 
>> This was my initial idea but I never got around to trying it out for 
>> real.
> 
> Yes my first approach was to try to implement it like this. But I could 
> not find or imagine a way for the server placing an RMI call to the 
> client without having to register an RMI object on the client side, as 
> the mechanism of listeners is a call-back from server to client. I also 
> consider this "back-connection" somewhat problematic.

OK, thanks for sharing this. In principle the client would not need to 
register the listener objects. It could just pass a RemoteListener 
reference to the server and the existing RMI connection would 
transparently handle the (now peer to peer) listener method invocations 
over the network without any need for registering or looking up a 
"listener service". Anyhow, I'm somewhat worried about the correct 
functioning of for example the RMI distributed garbage collector in such 
a setup. I'm not that much of an RMI guru to be fully confident about 
this idea. :-)

> Attached is a patch which fixes the issues regarding the plugin.xml, 
> copyright messages and the concurrent dependency.

Thanks.

One more thing before I commit your changes; the patch seems to be 
missing the o.a.j.rmi.observation.Queue class.

BR,

Jukka Zitting