Ben Sigelman

@adriancole you are sounding like `@RealDonaldTrump` with this "fake ref-counting" FUD! There is, in fact, a threadsafe count of references to an ActiveSpan or its Continuation. When the count gets...

@adriancole the reference count is incremented whenever a `Continuation` is `capture()`ed and decremented whenever an `ActiveSpan` is `deactivate()`d (or `close()`d, which is a simply a synonym for `deactivate()` that's provided...

(my recent comment from #134 applies here as well: basically, that this API seems like the right way to do what's being proposed, but that I am hesitant to add...

I'm ambivalent about this PR... I think it's "clean" in the sense of being extensible and clear, but inconvenient (in the sense that simple logging statements require three chained method...

:cool:, sounds good. Thanks @michaelsembwever

@yurishkuro I think that would be great... certainly would allow us to make faster progress since the stakes are lower. cc @hkshaw1990

@hkshaw1990 done: https://github.com/opentracing-contrib/go-observer/invitations

@hloos I think the right solution here will indeed involve a change to the default TLS span source impl... Did you try patching in the null check you suggested? If...

@scottmitch we are restricted to Java 1.6 features due to compatibility issues with Android (and other legacy stuff); I believe that prevents us from pursuing the approach above.

You're right -- github closed the issue automatically when that typo PR merged. Reopened.