Lalit Kumar Bhasin

`-DWITH_STL=ON` cmake option will set the `HAVE_CPP_STDLIB` pre-processor macro. You need to set this macro explicitly while compiling your test code using g++, something like below: `` g++ -DHAVE_CPP_STDLIB -o...

> BTW, I also had to switch to C++17 - with C++11 I got tons of compilation errors after I added -DHAVE_CPP_STDLIB. Yes, that's [expected](https://github.com/open-telemetry/opentelemetry-cpp/blob/b8fae39f81cfe079db84d7c5348b431b54580675/docs/dependencies.md#external-dependencies). It tries to use std::variant,...

> **My point here is that header file(s) could have some #if at the beginning to check C++ version and #error early. Agree, HAVE_CPP_STDLIB is the half-baked solution, not prioritized...

Have added my comments over the python issues, but I feel this needs wider discussion over the specs, to manage the cross-language span/trace portability.

Moving it out of RC, as the current specs is still not stable, and the complexities involved in implementing this.

That's correct, it's a valid bug. Unfortunately, we don't have any CI test to catch this error. For Linux - the nlohmann-json dependency is already installed through `apt`, and macOS...

> Downloading code manually from github sounds redundant with git submodules, which are already in place, so maybe (3) can be removed entirely ? We have source releases, so it...

> Implementation for (2) and (3) appear to be missing. Client authentication is missing. There is open issue for that - #389, and some work done in #938 > Should...

No problem. We can use this issue to track supporting secure options for OtlpHttpExporter.

> I feel supporting/maintaining two build systems is a little bit inconvenient Yes, I do agree it's inconvenient to maintain both these build systems. The decision to support both was...