Scott Schurr

Rebased to 1.9.2.

@ximinez, if I'm recalling correctly , the explicit `this` captures that you removed to make VS17 happy were added to prevent warnings from clang in C++20 mode. Should we consider...

I re-tested the most recent (squashed) version of this pull request on Mac/clang. I used boost 1.77.0. Worked great, no issues. Specifically, I did not see the RocksDBFactory.cpp issue that...

Unfortunately it will be hard to compare versions of clang. Clang on Mac uses its own versioning scheme. Here's what I get: ``` Apple clang version 11.0.0 (clang-1100.0.33.17) ```

Just to stick my nose in briefly, I'd like to discourage us from using modules in C++20. There are a number of issues that still need to be worked out,...

Rebased to 1.10.0-b1.

I thought I saw a fairly recent push to this pull request. So I tried building it. The previously observed build failures on macOS are still present.

@greg7mdp, I'm fine either way, either updating this pull request or creating a new one.

And, @greg7mdp, I tried building your branch locally. Yup, your version works way better on macOS than the previous version did. Thanks! I think I will wait until your version...

One person's cake is another person's poison. Indeed, another approach would be to see if a `yield()` instead of a `sleep_for()` returns the performance. But I _like_ the vector of...