Proof of Concept: Migrate to Cmake

[mchoo7]

This is not for review, I'm just suggesting possible implement of cmake version of lutok. Once there is a consensus, we can start review and improve.

I created a cmake prototype with help of AI to see how it works. Currently build and doxygen are tested, whereas the testing part needs cmake-version of atf. IMHO, cmake is way superior to autoconf and transition to cmake should be made at some point. Note that I bumped lutok version to 0.7.0 because <state.ipp> was merged into <state.hpp> (*.ipp isn't used often)

This project is now cmake only, and other projects using lutok needs to transition from autoconf to cmake.

Some other ideas to modernize lutok (also for atf and kyua):

• Run clang-format on all files. We don't see much changes to lutok these days and if we use "format file when there is a change" method, we will never be able to format all files.
• Safer and modern cpp. can be achieved in many ways (e.g. static analyzer)

[ngie-eign]

@mchoo7: I appreciate the effort !
First off: I need to check and see what the project's current stance on LLM use is. We were rejecting some submissions earlier last year, partly because of AI slop that some users were submitting, but also concerns over the licenses of the source material.
Second off: what LLM did you use and how did you use it? Was it Agentic AI, boilerplate produced by a product like Copilot/Windsurf, etc?

[mchoo7]

@mchoo7: I appreciate the effort ! First off: I need to check and see what the project's current stance on LLM use is. We were rejecting some submissions earlier last year, partly because of AI slop that some users were submitting, but also concerns over the licenses of the source material.

There is no "official" decision made yet. There were several past and ongoing efforts to codify LLM policies though. See D50650, D54366, and D54817.

TL;DR: The first proposal D50650 was about completely banning LLM usage like NetBSD and Gentoo, but there were many voices against it (even from some core members and senior committers). The only worry about LLM usage was license issue, but it seems trivial at this point (too many LLM-generated codes are already in-production and there hasn't been any noticeable litigation over it). We won't even have "Coauthored-by: " line in commit message to make sure that the only "author" is the contributor and not LLMs, and the coauthored-by label can blur the responsibility (LLMs can't have authorship).

The new proposal was submitted by a core member, so I think this is the project's final decision on LLM usage.

Second off: what LLM did you use and how did you use it? Was it Agentic AI, boilerplate produced by a product like Copilot/Windsurf, etc?

CmakeLists.txt is a boilerplate from CLion and .gitignore is from github/gitignore repository.

I used Claude code because I forgot Cmake syntax and going over documentation will take a long time while this is just a suggestion. I don't use "automatically accept all changes" (aka Shift+Tab) and I could find some mistakes while reviewing code it generated.