bosilca

We might need to add a dual mechanism as the one we use in OMPI for handling requests, they can be internally completed or externally completed, allowing us a window...

1. I don't understand how the C++ compiler expects an `extern "C"` function to be able to underact in any way with C++ exceptions. 2. If we add `noexcept` we...

I still don't see an answer to my first question. But one can always count on [stackoverflow](https://stackoverflow.com/questions/24362616/does-the-c-standard-mandate-that-c-linkage-functions-are-noexcept) for an answer. Short version, we need to add `noexcept` because `extern "C"`...

Triggering it is not the problem. The problem is that there are pending actions that are not tracked by the runtime directly, and that are accounted in the `pending actions`....

This scenario can indeed happen every time we reuse ID. The discussion was to move the complexity on the tool side, and match the event with the earliest completion event...

it is not only about how to generate the unique ID, but also how to store it in the code in order to reuse it later on.

That's exactly what I had in mind, extending #372 (now merged) to gracefully handle these cases, and remove the burden from the user.

Right, we don’t have the automatic fallback from the optimized path, because there is no guarantee that the supposedly matching end event on the same stream is the correct one....

Unfortunately, the proposed "solution" only works for the Intel compilers, all other compilers will choke with an unknown-pragma error. I fiddled a little around and it seems the culprit is...

Adding `CFLAGS=-fno-finite-math-only` solves the problem. Few additional issues: - ADD_COMPILER_FLAG_IF_SUPPORTED add the flags to all compilers, which requires the C and CXX compilers to be of the same type. I...