David Matson

@horenmar - Please let me know what you think of how this approached worked out.

@horenmar - it looks like you triggered a new validation run (thanks!). For this one: Linux builds (complex) / CMake configuration tests, clang++-10, C++14 Debug (pull_request) I'm guessing that's a...

@horenmar - thanks for the reply! With pipes, something needs to keep reading the pipe, or the OS buffer can get full and the writer will block. The only solutions...

@horenmar - one option is just to go with the [original plan](https://github.com/catchorg/Catch2/issues/2444#issuecomment-1146595534) and make the pipes implementation be just for Windows, and keep the existing temp file-based code for other...

> I am not convinced that a quarter-baked `jthread` polyfill is a good idea. Given that the only feature used from it is defaulting to join, I would rather see...

I'm not sure how much coverage there is for both variations of CATCH_CONFIG_NEW_CAPTURE (temp file vs pipe/thread). Ideally, we'd have: 1. Temp file on Linux (CATCH_CONFIG_NEW_CAPTURE but not threads) 2....

@ferdnyc - Thanks for your reply. If it's as simple as two lines, I have no objection. But I think there's the cost of having to fork the package definition...

And regarding this point: > vcpkg have explicitly said they don't support that, at least for C++ standard specifically, which may mean that Catch2 should just be considered incompatible with...

vcpkg solved this on their side: https://github.com/microsoft/vcpkg/pull/25019

@dimztimz - maybe that's best as a separate issue, since vcpkg was in the title of this one and that's now solved? Your proposed solution sounds good to me, but...