<div dir="ltr">After some more thought, I'm re-negging on this approach.<div><br></div><div>The underlying problem has to do entirely with dependent scoped beans within instance objects. I've raised a PR making a more appropriate spec change, clarifying more on the user side how they should work with them.</div><div><br></div><div>There's another section of the spec that I thought about adding this content to, and wouldn't be opposed to making that change.</div><div><br></div><div>John<br><br><div class="gmail_quote"><div dir="ltr">On Tue, May 3, 2016 at 1:20 PM John D. Ament <<a href="mailto:john.d.ament@gmail.com">john.d.ament@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">All,<div><br></div><div>I have opened a while ago - <a href="https://issues.jboss.org/browse/CDI-457" target="_blank">https://issues.jboss.org/browse/CDI-457</a> which was originally to add a disposable interface. The main driver was that if you're using an injection point like</div><div><br></div><div>@Inject</div><div>private Instance<MyDependentBean> beanInst;</div><div><br></div><div>doing beanInst.get() can cause leaking beans since there is no creational context. Upon looking at it further, there's some effort to do this in a safer way and leverage things like TransientReference.</div><div><br></div><div>This doesn't work for classes provided by external libraries, even the JDK itself (like String). I'd like to propose that the scope of this ticket be to allow TransientReference on producer fields/methods since right now its only allowed on types. </div><div><br></div><div>WDYT?</div></div><div dir="ltr"><div><br></div><div>John</div><div><br></div></div></blockquote></div></div></div>