friendlyanon

@satya-das has a WIP solution called [cib](https://github.com/satya-das/cib), which generates an hourglass interface similar to what was [described](https://www.slideshare.net/StefanusDuToit/cpp-con-2014-hourglass-interfaces-for-c-apis) by @sdutoit

"Optional" anything doesn't usually play nice with packaging, since with each option you double the configuration space. What you could do is make the CMake scripts offer just a static...

Once I get to the end of refactoring the CMake files and reduce the number of files the root one must touch, the required list of files can be narrowed...

It seems the idea has been floating around for some time and I have the time and willingness to look into this topic. From what I can tell, it would...

For reference, I have created [simdjson/ci-docker-images](https://github.com/simdjson/ci-docker-images) for this purpose.

The `#include`s would not change at all from a client's PoV who's using the CMake package. Let's assume that I have fplus installed to my system to the `/usr/include` location,...

Since I'm not aware of fplus being distributed in distro package managers based on the INSTALL.md, this might not be such a huge problem right now, but I wanted to...

This isn't really something the committee can change, people writing libraries should excercise some level of discipline and have knowledge about packages. I admit the resources to acquire this kind...

That's a lot of code to replicate what cache variables do already: ```cmake set(CMAKE_INSTALL_INCLUDEDIR "include/FunctionalPlus" CACHE PATH "") include(GNUInstallDirs) ``` `CMAKE_INSTALL_INCLUDEDIR` has [never](https://github.com/Kitware/CMake/blame/0f37cd228e9668d981ff75f3e1c63d4f716b7e57/Modules/GNUInstallDirs.cmake#L235) meant anything other than `include`, so the...

> Feel free to raise the minimum CMake version requirement if needed. I don't plan to and from the cursory look I took, it'd be wholly unnecessary. 3.8 is just...