geronimo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Lin Sun <linsun....@gmail.com>
Subject Re: RFC 66 and Geronimo
Date Thu, 08 Oct 2009 13:21:52 GMT
Hi,  see comments below.

Lin

On Thu, Oct 8, 2009 at 5:48 AM, Rick McGuire <rickmcg@gmail.com> wrote:

> The URL handler is a different stage of the processing.  That gets involved
> at the installation time of the bundle, and
> causes a transform of a standard WAR into a WAB, which then gets processed
> by the extender.  Once the WAB has
> been installed and started, that's when the WAB extender kicks in, and at
> that point, there's no opportunity to
> perform the transformation.

I think Geronimo has full control when the bundle/war is installed and
this can be triggered by geronimo's existing deployment mechanisms
(such as hot deployment, command line deployer or admin console
deployment).  I kinda think it is best to do this type of geronimo
specific conversion at the install time, because it would be too late
if we do it after the bundle is resolved/started.  The web extender
defined by RFC 66 is something that we must implement, but we could do
stuff beyond what is defined for the web extender. :)  What I am
proposing is to develop a geronimo url handler that can convert the
war file into the geronimo plugin (or whatever artifact that can be
deployed to geronimo's web server).  For example, when a war file is
installed, we can use the following as location identifier of
BundleContext.install(String loc).

webbundle:geronimo:file:///Users/linsun/test1.war?Web-ContextPath=/linsun

This will invoke the webbundle url handler which will convert the war
to a wab for Geronimo.  Then it will invoke the geronimo url handler
which will convert the wab to geronimo plugin (or whatever artifact
that can be deployed to geronimo's web server).

Then the bundle is installed and resolved.   Then the web extender
defined by rfc 66 kicks in and listens to the bundle events.

> "Should" items are there for a reason and at least an attempt should be made
> to satisfy the language of the specification.
> If it turns out to be impossible, so be it, but the possibility of fully
> implementing that needs to be made.

Right i just want to point that out. :)

Mime
View raw message