Feature: CMake Build Runners for All Supported Platforms

[thearchivalone]

Is your feature request related to a problem? Please describe.

Having both the QBS and CMake build runners for supported platforms would allow for proper testing between both build systems. If there’s a way to do a checksum test between the artifacts produced from the platform builds, that would also allow for finding possible build configuration errors. (Not sure how viable or easy this would be yet)

Describe the solution you'd like

Implement this into the current GitHub build workflow and decide on overall equal build configurations for both (which is what’s already in place with the build scripts used).

Describe alternatives you've considered

I haven’t looked into direct alternatives.

Additional context

This would increase the total build times overall but would also allow for future bugs in PRs to be found more easily.

[thearchivalone]

For FreeBSD, the core build tool for native applications is gmake instead of make but I believe ninja is the recommended way for all platforms. I'll look into it further but I think having a single build tool to go with CMake set at a version that all platforms support would make testing and updating the runner consistent across the board. It would also make changing to a different one if needed a simpler process. https://ninja-build.org/

I personally love this tool but it's not CMake friendly (afaik) and may not be compatible with all platforms: https://github.com/facebook/buck2 . It's also a bit of a pain to get setup for those that haven't used it before.

[thearchivalone]

So my plan is to start working on this on my fork first and testing it out with this unofficial runner: https://github.com/vmactions/freebsd-vm . That way changing the build system doesn't break the code sanity here upstream. I'll keep you posted and updated on the progress but once it's stable, tested and working, I'll open a PR with no expectation of you rushing to merge it immediately. Breaking your build system by adding new runners without testing is insane. 😅

[thearchivalone]

I was hoping to have my first successful compile with the freebsd-vm fully ready before going to work today but I found a nice little quirk where when building in parallel, certain moc headers won't always generate before they're needed. Running as a single job takes a long time on the initial build but is always guaranteed to generate properly if everything else is right. I may have a workaround for that so the builds can be faster (not 30 minutes like it was in this test) but I'll need to experiment further before sending the PR.

No artifacts upload yet but here's a successful build for FreeBSD 14.1: https://github.com/bedwardly-down/thunder-bsd/actions/runs/9827166404/job/27129553905

[thearchivalone]

We have an Artifact and a bandage fix for now: https://github.com/bedwardly-down/thunder-bsd/actions/runs/9827480043