mirror of
https://github.com/OrcaSlicer/OrcaSlicer.git
synced 2026-09-16 21:42:43 +00:00
Kris Austin
46180c3f54
* build: clear 2 warnings - a precedence bug and an arm64-only pragma Both were found by promoting every warning to an error across the CI matrix. Neither is reported by clang-cl on Windows x64, which is the configuration the #15374 inventory measures. LineSplit.hpp reserved with path.size() + closed ? 1 : 0. Addition binds tighter than ?:, so that parses as (path.size() + closed) ? 1 : 0, and the function returns early when path is empty, so the condition is always true and the reserve is always 1. The vector then grows by reallocation instead of reserving once. Output is unaffected, since reserve only sets capacity. Reported by Clang on Linux, macOS and Flatpak; GCC does not diagnose it. Int128.hpp declared #pragma intrinsic(_mul128) under _WIN64, which is defined on Windows arm64 as well, where that x64 intrinsic does not exist. The call site at line 190 is already guarded on _M_X64 and carries a comment saying ARM64 has no _mul128, so the pragma now uses the same guard. x64 is unchanged because _M_X64 is defined there. * build: clear 1 warning - CLI error label prints 1 instead of a name construct_assemble_list is a function, so streaming it converts the function pointer to bool. When that catch block fires the CLI prints "1: <message>". This line was already fixed in #5963 and came back in the wholesale revert of that PR two weeks later, which was reverting an auto-orientation regression somewhere in its 184 files. The string is restored exactly as it was merged then.