feat(CI): Windows MSVC and GCC build

[liamHowatt]

Description of the feature or fix

See #5904 and #5978

Add a CI test for building on Windows. In particular, build the Windows-specific files from #5313.

Notes

• Update the Documentation if needed.
• Add Examples if relevant.
• Add Tests if applicable.
• If you added new options to lv_conf_template.h run lv_conf_internal_gen.py and update Kconfig.
• Run scripts/code-format.py (astyle version v3.4.12 needs to be installed) and follow the Code Conventions.
• Mark the Pull request as Draft while you are working on the first version, and mark is as Ready when it's ready for review.
• When changes were requested, re-request review to notify the maintainers.
• Help us to review this Pull Request! Anyone can approve or request changes.

[kisvegabor]

Very nice! It would be great to reuse the CMake files for MSVC too, but if there is more Windows-ish way we can choose that too.

@MouriNaruto could you take a look at the Windows driver related changes?

[MouriNaruto]

@kisvegabor I have read the changes. I think @liamHowatt should move the device window message handler declarations back because these are private interfaces.

For LVGL Windows backend, I try my best to maintain the ABI compatibility. So, I don't hope to see making some private interfaces to be public. (The lv_windows_input.h and lv_windows_display.h only contain the public API. The lv_windows_context.h only contains the context-related private APIs.)

Kenji Mouri

[liamHowatt]

Thanks for the feedback @MouriNaruto. In hindsight I see that those handlers should not be the header as you say. I originally did it to fix a CI warning.

https://github.com/liamHowatt/lvgl/actions/runs/8527506893/job/23359087388

Error: D:/a/lvgl/lvgl/src/drivers/windows/lv_windows_input.c:283:6: error: no previous prototype for 'lv_windows_pointer_device_window_message_handler' [-Werror=missing-prototypes]
  283 | bool lv_windows_pointer_device_window_message_handler(
      |      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Generally it's good to only-declare-once. Recently we've been moving private cross-file declarations to *_private.h headers #5913. What do you think of my latest commit(s) as a potential compromise?

[kisvegabor]

If @MouriNaruto agrees with having lv_windows_input_private.h, please add it to lvgl_private.h too.

[MouriNaruto]

@kisvegabor I'm agree with having lv_windows_input_private.h.

[XuNeo]

Nice work!

[MouriNaruto]

@liamHowatt I think you should do some macro checks for GCC-specific progma to keep the MSVC compatibility.

Kenji Mouri

[kisvegabor]

Will you migrate to MSVC still in this PR?

[liamHowatt]

I think you should do some macro checks for GCC-specific progma to keep the MSVC compatibility.

Yep, good idea.

Will you migrate to MSVC still in this PR?

That wasn't the intent of this PR but yes I will. It is important.

[kisvegabor]

Will you migrate to MSVC still in this PR?

That wasn't the intent of this PR but yes I will. It is important.

Yes, I think MinGW is not so interesting compared to MSVC.

[liamHowatt]

There is now an MSVC build workflow.

• Using /W3 warning flag to enable all "production quality" warnings. Not using /W4 because it gives too many Spectre vulnerability mitigation warnings.
• Manually set warning C4013 to be treated as an error. It is the "implicit function declaration" warning.
• The warnings from chore: change position and size parameters to size_t to avoid 64-bit system compiler warnings #5873 are being emitted.
• I cleaned up some concerning warnings but left most of them alone for after 5873 gets merged. There are a lot of integer conversion size/sign warnings as well as signed/unsigned comparison warnings. build logs
• ASM disabled for MSVC in CMake

[kisvegabor]

Thank you!

At some point it would be nice to lower the number of the remaining warnings, but at least now we know that LVGL builds on MSVC too! 🎉