[weld-dev] standard servlet context attribute for the BeanManager

[Dan Allen]

On Wed, Mar 24, 2010 at 11:59 AM, David Allen <drallendc at gmail.com> wrote:

> Am Mittwoch, den 24.03.2010, 11:36 -0400 schrieb Dan Allen:
> > On Wed, Mar 24, 2010 at 11:24 AM, David Allen <drallendc at gmail.com>
> > wrote:
> >         Am Mittwoch, den 24.03.2010, 11:11 -0400 schrieb Dan Allen:
> >         > On Wed, Mar 24, 2010 at 3:38 AM, Emmanuel Bernard
> >         > <emmanuel at hibernate.org> wrote:
> >         >         But how do you solve that in SE? Or even in a web
> >         environment,
> >         >         more and more web environment are getting away from
> >         the
> >         >         servlet API imposed constraint.
> >         >
> >         >
> >         > Sigh. And that's why this problem is more difficult than it
> >         has to be.
> >         > Because we can't even handle the 80-90% case.
> >         >
> >         >
> >         > But you did get me thinking. Maybe the way to address the
> >         problem is
> >         > to see this more as an integration concern for the SPI. So
> >         if I'm
> >         > running in a servlet environment (whether it be a servlet
> >         container or
> >         > Java EE), then I'm providing an implementation that
> >         standardizes on
> >         > the servlet context. But we can define other mappings for
> >         other known
> >         > environments, or unknown environments can supply the SPI
> >         impl.
> >         >
> >         >
> >         > Btw, it's not like JNDI is all that portable to SE either ;)
> >
> >
> >         Actually JNDI is quite simple and can be used in any Java
> >         application.
> >         It does not necessarily require a server.  It is just an API
> >         and a
> >         simple implementation can always be used to provide the
> >         necessary
> >         functionality within an SE.
> >
> >         Here's one example:
> >          http://www.osjava.org/simple-jndi/index.html
> >
> >
> > What I meant by portable is that an extension can assume it is
> > present. We can only assume it is present in Java EE, because it is
> > only required there. Servlet containers, while they do have JNDI,
> > don't have a writable java:comp/env JNDI namespace, so that's pretty
> > much the same as it not being there. That's the whole reason we are in
> > this fix.
>
> Yeah, this might be true in general; however, we do have Weld working in
> Tomcat, and I believe it is entered into JNDI there as well.  It took
> some work to do that, though, since that space is normally read-only at
> runtime.  And actually that was done under the old java:app namespace.
> Maybe this is even more difficult under java:comp/env.
>
> My understanding was that onus is on the developer to include the
BeanManager definition in context.xml (and similar for Jetty). To me, that's
just an unnecessary requirement. I'm being tough because a lot of end
developers are going to see this requirement and be like "forget it, too
much work".

-Dan

-- 
Dan Allen
Senior Software Engineer, Red Hat | Author of Seam in Action
Registered Linux User #231597

http://mojavelinux.com
http://mojavelinux.com/seaminaction
http://www.google.com/profiles/dan.j.allen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/weld-dev/attachments/20100324/0cba8f54/attachment-0001.html