Background

[ryanmrichard]

Is this pull request associated with an issue(s)?
No

Description
This PR adds some background to the CMaize documentation.

TODOs

• Look for more similar projects
• Clean up Statement of Need.

[ryanmrichard]

For tomorrow:

https://github.com/ixm-one/legacy
https://github.com/cpm-cmake/CPM.cmake
https://github.com/iauns/cpm
https://github.com/vector-of-bool/pmm

[ryanmrichard]

@AutonomicPerfectionist I have a weird bug. If you pull this branch and try to configure it (cmake -H. -Bbuild should suffice), you should get infinite recursion in CMakeTest's message override. The offending message call is in CMaize/tests/docs/CMakeLists.txt. I checked it in with _message, but I also tried without the _ and ran into the same problem.

The actual message call seems fine to me (right now it's literally a string because I took the $ off of the variable I was trying to debug). So the override should just be calling the native message command...

[ryanmrichard]

FWIW, I realize this bug is better reported in the CMakeTest repo, but I'm not sure what combination of interactions is causing the bug in order to piece together a MWE.

[AutonomicPerfectionist]

Infinite recursion in that override is a recurring (heh) problem, it usually happens because a fatal error is thrown somewhere along the exception handling route, which then starts the recursion over again. I'll take a look in a bit

[AutonomicPerfectionist]

Well this is a new one, somehow the overrides are being included more than once even though there's an include_guard, so the underscore function is overridden as well. Commenting out the include("cmake_test/cmake_test") in tests/CMakeLists.txt fixes it, but obviously that's not a real solution. Going to have to do some more digging to find A) where else cmake_test is being included and B) why the include_guard isn't working

[AutonomicPerfectionist]

@ryanmrichard a small fix in CMakeTest appears to solve it at first glance, probably should do some more tests before merging it though

[ryanmrichard]

@AutonomicPerfectionist Thanks for fixing that so fast! As for more testing, it looks like it fixes this problem; so I'm fine merging it and revisiting the issue if something else comes up.

[ryanmrichard]

@zachcran I'm going to freeze this PR here. It quickly got away from me. As I think you'll see this PR really will be the bulk of the paper's background. In preparing that, I really wanted to showcase the simplicity of CMaize vs. CMake. For this reason I wanted code snippets. This led me to needing to unit test the documentation examples to ensure I wasn't missing something (they really did work as advertised). This then led to me basically re-inventing some of what I think underlies your integration tests, though they're run out of this repo in a hopefully platform agnostic way. Basically:

• CMake/CTest calls CMake in scripting mode.
• In scripting mode we call CMake in build mode to actually build a sub-project.
• After building (and potentially installing), still in the original scripting mode call, we again run CMake in scripting mode, this time to run a script which checks the output of the build.

[ryanmrichard]

It looks like me defaulting the C++ version to 98 (when nothing is selected broke things. I'm going to try going back to the old behavior (no value) and skipping the relevant line in _set_compile_features when it's empty (which was my original error).

[zachcran]

@ryanmrichard Sorry for the delay. I'll take a look at this over the weekend!

[github-actions]

🚀 [bumpr] Bumped! New version:v0.1.11 Changes:v0.1.10...v0.1.11