Ryan Burn

> In rdc case, objects from deps should always propagate for dlink, and in non-rdc case, objects does not propagate I didn't put together an issue yet. But I was...

> That is because cuda_objects only provides a CcInfo with only compilation_context, but no linking context. As the original design, cuda_objects is just compilation actions, whose actifacts are to be...

We did something similar with the LightStep tracer, where you can pass in an [std::function](https://github.com/lightstep/lightstep-tracer-cpp/blob/master/include/lightstep/tracer.h#L78) to capture and redirect any logging output.

Thanks @jmcomets - I'd be glad to have this functionality added in. I'll try to review this soon.

Could you add some coverage for this into https://github.com/rnburn/zipkin-cpp-opentracing/blob/master/zipkin_opentracing/test/ot_tracer_test.cc#L129?

@jmcomets I actually had setValue use non-string values at one point, but I changed it to strings because the non-strings broke jaeger's zipkin compatibility mode. According to https://github.com/openzipkin/zipkin/issues/939#issue-131157141, non-string binary...

Great! I'd welcome PRs to support customizing the configuration more and adding dropped span metrics. A couple of thoughts on the proposal: > 1. Add member functions to the reporterImpl...

> I chose this method to allow for dynamic run time adjustment of the configuration. In some scenarios these values are unknown at initialization time and need to be adjusted...

> I could do the instantiation config parameters as one change, expose the OtTracer as a second, and everything else as the third. That sounds good.

Thanks, could you put in a PR for this. Also, is there any documentation about this way of using the "expect:" header you can point me to?