Hi Tim-
Thanks for replying. I assumed that was likely the case. The manifest class
path approach works fine as expected, so no big deal. I'll open an RFE and
you all can toss the idea around.
Best regards,
Dave
On Wed, Sep 24, 2008 at 2:48 PM, <glassfish_at_javadesktop.org> wrote:
> Hi, Dave.
>
> What you describe is not supported currently.
>
> It's an interesting idea; go ahead and open an RFE for it.
>
> The Java EE spec does not directly address directory deployment, even
> though it's a common feature in app server implementations. The spec does
> say this, on the subject of the lib directory (or library-directory setting
> if present):
>
> "All files in this directory (but not subdirectories) with a .jar extension
> must
> be made available to all components packaged in the EAR file, including
> application clients. "
>
> Because directory deployment is already outside the spec (although not in
> contradiction to the spec) it might not be too much of a stretch to allow
> directories (as well as JARs) in the lib directory to be included in the
> class path for directory-deployed apps only. But that would, strictly
> speaking, go against the spec as to exactly what in the lib directory should
> be available to the application.
>
> Another approach you might consider for the classes that you control would
> be to deploy them in directories in the EAR (not under lib) and then refer
> to those directories in the Class-Path of the manifest of the components
> that depend on those classes. This falls within the current spec and,
> therefore, the current implementation.
>
> - Tim (one of those deployment people Vince referred to)
> [Message sent by forum member 'tjquinn' (tjquinn)]
>
> http://forums.java.net/jive/thread.jspa?messageID=301373
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>
>
--
==========
Dave Muirhead
dmuirhead_at_gmail.com