SCons: Colorize warnings/errors during generation

[Repiteo]

This change adds the ability to substitute print with methods.print_warning or methods.print_error, prepending the messages with WARNING: /ERROR: and colorizing them as yellow/red respectively. The messages will also be routed to stderr instead of stdout, allowing users to silence stdout without missing on any crucial information if a build fails or otherwise stops. non-atty environments won't benefit from colorization, but will still enjoy prefix & stderr benefits.

EDIT: Supersedes #88074

Bugsquad edit:

• Closes Make warnings/notes in build scripts colored godot-proposals#9291

[Riteo]

Holy moly you rock!

Took a look and the diff looks fine!

[Calinou]

Does this supersede #88074? It includes some not changes not made in this PR.

[Repiteo]

I don't think it supercedes that PR entirety; though, it probably would need a different title if left as-is

[Riteo]

Oops, I forgot about the older PR, I'm sorry @Calinou.

This is a bit of an unfortunate situation. How should we handle it?

They're both great PRs, with the same main change but different details.

[Repiteo]

Did some cleanup work; now it supersedes #88074! The only bits I didn't bring over were print_info because it felt a little redundant & the "Enable ANSI escape code support on Windows 10 and later" section as that issue appears to have been fixed in Python 3.7 which is good enough for me.

[Calinou]

Sounds good to me. We should make sure to update the Compiling on Windows documentation to require Python 3.7 though, if we start requiring it de facto for proper display of errors/warnings on Windows.

[Repiteo]

Ehhh, if we gotta go through that trouble I'll just add that bit of code after all; I'll throw in a note that it should be removed when the minimum version is >=3.7

[Repiteo]

Hmm, I might've misread the original post; I don't think it's necessarily limited to 3.7 after all. And yet, I've never had any issues with displaying colored text & I use Windows 11, so I honestly have no clue what circumstances cause that bug to occur. Either way, no real harm done in including it.

[tiloc]

Sounds good to me. We should make sure to update the Compiling on Windows documentation to require Python 3.7 though, if we start requiring it de facto for proper display of errors/warnings on Windows.

Python 3.7 reached end-of-life 10 months ago. 3.8 and later still get security fixes.

[Calinou]

Tested locally (Windows 11 23H2, MSVC 2022, Python 3.12.3), it works as expected. Code looks good to me.

[Calinou]

Compilation output on CI looks strange with lots of [Initial build], which I don't see on other PRs' CI logs: https://github.com/godotengine/godot/actions/runs/8868984410/job/24349194236?pr=91220

Perhaps we should use progress=no on CI either way, as I don't think logging progress on CI is that useful.

[Repiteo]

I double-checked the PR this supersedes, and that error was happening there too. Seems like enabling progress reporting for CI is what caused it, as reverting that fixed the issue.

[akien-mga]

Thanks!

[Anutrix]

This MR broke use_mingw flag on windows:

PS D:\Repo\godot> scons use_mingw=yes
scons: Reading SConscript files ...
Automatically detected platform: windows
Auto-detected 24 CPU cores available for build parallelism. Using 23 cores by default. You can override it with the -j argument.
Using MinGW, arch x86_64
Building for platform "windows", architecture "x86_64", target "editor".
AttributeError: 'str' object has no attribute 'insert':
  File "D:\Repo\godot\SConstruct", line 971:
    methods.no_verbose(env)
  File "D:\Repo\godot\methods.py", line 671:
    env.Append(SHLINKCOMSTR=link_shared_library_message)
  File "C:\Users\<username>\AppData\Roaming\Python\Python312\site-packages\SCons\Environment.py", line 1458:
    val.insert(0, orig)

Note: Compilation proceeds when verbose flag is given.