Optimize to_string

[YexuanXiao]

C++23 added resize_and_overwrite to avoid the problem of initializing the entire string to zero during resize. winrt::to_hstring(std::string_view) can take advantage of this feature.

Additionally, using value ? L"true" : L"false" instead of returning in two separate if branches allows the compiler to optimize better, avoiding the generation of two constructor calls, see https://godbolt.org/z/xeY5K4vMh.

[Copilot]

Pull request overview

This PR optimizes string conversion utilities in winrt by reducing unnecessary initialization work when converting wide strings to UTF-8, and by simplifying boolean to_hstring conversion to enable better compiler optimization.

Changes:

• Simplifies to_hstring(bool) to a single ternary-based return expression.
• Uses C++23 std::string::resize_and_overwrite (when available) in to_string(std::wstring_view) to avoid pre-initializing the entire buffer.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

Suppressed comments (1)

strings/base_string.h:721

• The new resize_and_overwrite block uses tab indentation (inconsistent with the surrounding 4-space indentation) and ignores the size passed by resize_and_overwrite (second lambda parameter), instead relying on the outer size variable. Using the provided count parameter both improves readability and ensures the conversion uses the buffer size actually provided by the library.

    	std::string result;
    	result.resize_and_overwrite(size, [&](char* buffer, std::size_t) -> std::size_t
    		{
    			WINRT_VERIFY_(size, WINRT_IMPL_WideCharToMultiByte(65001 /*CP_UTF8*/, 0, value.data(), static_cast<std::int32_t>(value.size()), buffer, size, nullptr, nullptr));
    			return size;

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

Suppressed comments (3)

strings/base_string.h:720

• The new C++23 resize_and_overwrite block uses tab indentation and inconsistent spacing compared to the surrounding code (which uses spaces). This makes the file formatting inconsistent and harder to review/maintain.

    	std::string result;
    	result.resize_and_overwrite(size, [&](char* buffer, std::size_t) -> std::size_t
    		{
    			auto bytes_written = WINRT_IMPL_WideCharToMultiByte(65001 /*CP_UTF8*/, 0, value.data(), static_cast<std::int32_t>(value.size()), buffer, size, nullptr, nullptr);
                WINRT_VERIFY_(size, bytes_written);

strings/base_string.h:725

• In the non-resize_and_overwrite branch, WINRT_VERIFY_ is a no-op in release builds, so if WideCharToMultiByte unexpectedly fails the function will return a string full of '?' characters. The C++23 branch would return an empty string (via returning 0 from the overwrite op), so behavior becomes inconsistent across standard library versions. Consider explicitly handling the failure/mismatch here as well and returning {}.

        std::string result(size, '?');
        WINRT_VERIFY_(size, WINRT_IMPL_WideCharToMultiByte(65001 /*CP_UTF8*/, 0, value.data(), static_cast<std::int32_t>(value.size()), result.data(), size, nullptr, nullptr));

strings/base_string.h:715

• PR description says the resize_and_overwrite optimization is for winrt::to_hstring(std::string_view), but the actual change is in to_string(std::wstring_view). Please align the description (or adjust the implementation if to_hstring(std::string_view) was the intended target) so reviewers/users understand what’s being optimized.

#if defined(__cpp_lib_string_resize_and_overwrite) && __cpp_lib_string_resize_and_overwrite >= 202110L

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

Suppressed comments (1)

strings/base_string.h:722

• WINRT_VERIFY_ compiles to a no-op in non-debug builds (strings/base_macros.h:15). In that configuration, returning 0 when bytes_written != size would silently truncate the result to an empty string. Returning the actual bytes_written better matches std::string::resize_and_overwrite’s contract and avoids dropping partially written output. Also, these added lines use tab indentation while the surrounding code uses spaces.

    	result.resize_and_overwrite(size, [&](char* buffer, std::size_t) -> std::size_t
    		{
    			auto bytes_written = WINRT_IMPL_WideCharToMultiByte(65001 /*CP_UTF8*/, 0, value.data(), static_cast<std::int32_t>(value.size()), buffer, size, nullptr, nullptr);
    			WINRT_VERIFY_(size, bytes_written);
    			return bytes_written == size ? size : 0;