Mailing-List: contact dev-help@geronimo.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@geronimo.apache.org
Received-SPF: pass (asf.osuosl.org: local policy)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
  s=s1024; d=yahoo.com;
  h=Received:Mime-Version:In-Reply-To:References:Content-Type:Message-Id:Content-Transfer-Encoding:From:Subject:Date:To:X-Mailer;
  b=TwrApqH6nf9UNEW/yyZoItiVtWHWJ/Er2c/wW3fEsIwJZ+8BKWXpaZFsuhA+PvFyVULjs9+Kvr/CQIOd3QhcA9b2iqB2VT3+y+wrP5mI8kBStOkbdTL4GWmvmj626xn4PEH7lEsIfagyQgq/egI8KrsDsA6XDIZCK6jsavJNAoY=
  ;
Mime-Version: 1.0 (Apple Message framework v746.2)
In-Reply-To: <44058BFB.7030801@optusnet.com.au>
References: <44058BFB.7030801@optusnet.com.au>
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
Message-Id: <270654D5-E8FB-4C4A-A9D3-90544AAC3154@yahoo.com>
Content-Transfer-Encoding: 7bit
From: David Jencks <david_jencks@yahoo.com>
Subject: Re: ModuleBuilder - add initENC after addGBeans
Date: Wed, 1 Mar 2006 07:27:27 -0800
To: dev@geronimo.apache.org


On Mar 1, 2006, at 3:56 AM, Gianny Damour wrote:

> Hi,
>
> I think that we need to split ModuleBuilder.addGBeans into two  
> methods: addGBeans and initENC. addGBeans implementations perform  
> GBean registrations as per the current approach. initENC is invoked  
> after the addGBeans phase and implementations use this callback to  
> build the ENC.
>
> The issue with the current implementation is that it is impossible  
> to bind a GBean reference to the ENC if the referenced GBean is  
> defined by a module which has not yet been processed by addGBeans.  
> For instance, if a module A references a GBean added by a module B  
> and if module A is processed before module B, then it is impossible  
> to locate the referenced GBean as it has not yet been added to the  
> registry.
>
> If there is no objection, I will start to work on it in the next  
> couple of days.

I would rather see us reduce the number of module builder phases than  
increase them :-)  So far we have dealt with this problem by  
registering a gbean data for anything that could possibly be  
referenced in the ENC during the initContext phase.  I'm also worried  
that changes of this nature could make merging the configid/1.1  
changes into trunk almost impossible.

What feature are you implementing that needs this that doesn't work  
with the current approach?  Could you wait with this until the  
configid work is merged to trunk?

thanks
david jencks

>
> Thanks,
> Gianny
>
>