MSVC 2022 build fails: C compound literals in C++ sources (C4576) + narrowing init in ScoreOptions (C2397)

[BeneOblivion-byte]

Environment

• OS: Windows (MSVC toolchain)
• Compiler: MSVC 2022
• Generator: Ninja
• CMake: (any recent version; I used CMake-based build integrating this repo as sources)

Problem 1: Non-standard C compound literal syntax used in C++ ((Vec3){...}) triggers MSVC error C4576

MSVC rejects C compound literals in C++ mode. There are multiple occurrences like:

return (Vec3){r, g, b};

This produces (example):

• cam/hct_solver.cc:281: error: C4576: a parenthesized type followed by an initializer list is a non-standard explicit type conversion syntax

Vec3 is defined as an aggregate type:

struct Vec3 {
  double a = 0.0;
  double b = 0.0;
  double c = 0.0;
};

Possible fix

Replace C compound literals with standard C++ aggregate initialization, e.g.

• return (Vec3){r, g, b}; -> return Vec3{r, g, b};
• return (Vec3){-1.0, -1.0, -1.0}; -> return Vec3{-1.0, -1.0, -1.0};

This should be portable across MSVC/Clang/GCC.

Problem 2: Narrowing conversion in ScoreOptions default member initializer triggers MSVC error C2397

In cpp/score/score.h:

struct ScoreOptions {
  size_t desired = 4;
  int fallback_color_argb = 0xff4285f4;  // Google Blue.
  bool filter = true;
};

MSVC reports:

• cpp/score/score.h:39: error: C2397: conversion from 'unsigned int' to 'int' requires a narrowing conversion

Possible fix

Use an unsigned 32-bit type for ARGB (or use the existing Argb typedef if applicable), e.g.

• Argb fallback_color_argb = 0xff4285f4u;
  or
• uint32_t fallback_color_argb = 0xff4285f4u;

(Using int for ARGB values that may exceed INT_MAX seems problematic on its own.)

Question

Is MSVC intended to be supported for the C++ implementation? If yes, I’m happy to send a PR that replaces (Vec3){...} with Vec3{...} and adjusts the fallback_color_argb type to avoid narrowing.